2019-05-27 08:55:01 +02:00
/* SPDX-License-Identifier: GPL-2.0-or-later */
2005-09-26 16:04:21 +10:00
/ *
* PowerPC v e r s i o n
* Copyright ( C ) 1 9 9 5 - 1 9 9 6 G a r y T h o m a s ( g d t @linuxppc.org)
*
* Rewritten b y C o r t D o u g a n ( c o r t @cs.nmt.edu) for PReP
* Copyright ( C ) 1 9 9 6 C o r t D o u g a n < c o r t @cs.nmt.edu>
* Adapted f o r P o w e r M a c i n t o s h b y P a u l M a c k e r r a s .
* Low- l e v e l e x c e p t i o n h a n d l e r s a n d M M U s u p p o r t
* rewritten b y P a u l M a c k e r r a s .
* Copyright ( C ) 1 9 9 6 P a u l M a c k e r r a s .
*
* Adapted f o r 6 4 b i t P o w e r P C b y D a v e E n g e b r e t s e n , P e t e r B e r g n e r , a n d
* Mike C o r r i g a n { e n g e b r e t | b e r g n e r | m i k e j c } @us.ibm.com
*
2009-06-02 21:17:38 +00:00
* This f i l e c o n t a i n s t h e e n t r y p o i n t f o r t h e 6 4 - b i t k e r n e l a l o n g
* with s o m e e a r l y i n i t i a l i z a t i o n c o d e c o m m o n t o a l l 6 4 - b i t p o w e r p c
* variants.
2005-09-26 16:04:21 +10:00
* /
2022-11-14 23:27:41 +05:30
# include < l i n u x / l i n k a g e . h >
2005-09-26 16:04:21 +10:00
# include < l i n u x / t h r e a d s . h >
2014-01-09 00:44:29 -05:00
# include < l i n u x / i n i t . h >
2005-10-10 14:01:07 +10:00
# include < a s m / r e g . h >
2005-09-26 16:04:21 +10:00
# include < a s m / p a g e . h >
# include < a s m / m m u . h >
# include < a s m / p p c _ a s m . h >
2016-09-28 11:31:48 +10:00
# include < a s m / h e a d - 6 4 . h >
2005-09-26 16:04:21 +10:00
# include < a s m / a s m - o f f s e t s . h >
# include < a s m / b u g . h >
# include < a s m / c p u t a b l e . h >
# include < a s m / s e t u p . h >
# include < a s m / h v c a l l . h >
[PATCH] powerpc: Merge thread_info.h
Merge ppc32 and ppc64 versions of thread_info.h. They were pretty
similar already, the chief changes are:
- Instead of inline asm to implement current_thread_info(),
which needs to be different for ppc32 and ppc64, we use C with an
asm("r1") register variable. gcc turns it into the same asm as we
used to have for both platforms.
- We replace ppc32's 'local_flags' with the ppc64
'syscall_noerror' field. The noerror flag was in fact the only thing
in the local_flags field anyway, so the ppc64 approach is simpler, and
means we only need a load-immediate/store instead of load/mask/store
when clearing the flag.
- In readiness for 64k pages, when THREAD_SIZE will be less
than a page, ppc64 used kmalloc() rather than get_free_pages() to
allocate the kernel stack. With this patch we do the same for ppc32,
since there's no strong reason not to.
- For ppc64, we no longer export THREAD_SHIFT and THREAD_SIZE
via asm-offsets, thread_info.h can now be safely included in asm, as
on ppc32.
Built and booted on G4 Powerbook (ARCH=ppc and ARCH=powerpc) and
Power5 (ARCH=ppc64 and ARCH=powerpc).
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-21 15:45:50 +10:00
# include < a s m / t h r e a d _ i n f o . h >
2006-09-25 18:19:00 +10:00
# include < a s m / f i r m w a r e . h >
2007-08-20 14:58:36 +10:00
# include < a s m / p a g e _ 6 4 . h >
2008-04-17 14:35:01 +10:00
# include < a s m / i r q f l a g s . h >
2010-04-16 00:11:32 +02:00
# include < a s m / k v m _ b o o k 3 s _ a s m . h >
2010-11-18 15:06:17 +00:00
# include < a s m / p t r a c e . h >
powerpc: Rework lazy-interrupt handling
The current implementation of lazy interrupts handling has some
issues that this tries to address.
We don't do the various workarounds we need to do when re-enabling
interrupts in some cases such as when returning from an interrupt
and thus we may still lose or get delayed decrementer or doorbell
interrupts.
The current scheme also makes it much harder to handle the external
"edge" interrupts provided by some BookE processors when using the
EPR facility (External Proxy) and the Freescale Hypervisor.
Additionally, we tend to keep interrupts hard disabled in a number
of cases, such as decrementer interrupts, external interrupts, or
when a masked decrementer interrupt is pending. This is sub-optimal.
This is an attempt at fixing it all in one go by reworking the way
we do the lazy interrupt disabling from the ground up.
The base idea is to replace the "hard_enabled" field with a
"irq_happened" field in which we store a bit mask of what interrupt
occurred while soft-disabled.
When re-enabling, either via arch_local_irq_restore() or when returning
from an interrupt, we can now decide what to do by testing bits in that
field.
We then implement replaying of the missed interrupts either by
re-using the existing exception frame (in exception exit case) or via
the creation of a new one from an assembly trampoline (in the
arch_local_irq_enable case).
This removes the need to play with the decrementer to try to create
fake interrupts, among others.
In addition, this adds a few refinements:
- We no longer hard disable decrementer interrupts that occur
while soft-disabled. We now simply bump the decrementer back to max
(on BookS) or leave it stopped (on BookE) and continue with hard interrupts
enabled, which means that we'll potentially get better sample quality from
performance monitor interrupts.
- Timer, decrementer and doorbell interrupts now hard-enable
shortly after removing the source of the interrupt, which means
they no longer run entirely hard disabled. Again, this will improve
perf sample quality.
- On Book3E 64-bit, we now make the performance monitor interrupt
act as an NMI like Book3S (the necessary C code for that to work
appear to already be present in the FSL perf code, notably calling
nmi_enter instead of irq_enter). (This also fixes a bug where BookE
perfmon interrupts could clobber r14 ... oops)
- We could make "masked" decrementer interrupts act as NMIs when doing
timer-based perf sampling to improve the sample quality.
Signed-off-by-yet: Benjamin Herrenschmidt <benh@kernel.crashing.org>
---
v2:
- Add hard-enable to decrementer, timer and doorbells
- Fix CR clobber in masked irq handling on BookE
- Make embedded perf interrupt act as an NMI
- Add a PACA_HAPPENED_EE_EDGE for use by FSL if they want
to retrigger an interrupt without preventing hard-enable
v3:
- Fix or vs. ori bug on Book3E
- Fix enabling of interrupts for some exceptions on Book3E
v4:
- Fix resend of doorbells on return from interrupt on Book3E
v5:
- Rebased on top of my latest series, which involves some significant
rework of some aspects of the patch.
v6:
- 32-bit compile fix
- more compile fixes with various .config combos
- factor out the asm code to soft-disable interrupts
- remove the C wrapper around preempt_schedule_irq
v7:
- Fix a bug with hard irq state tracking on native power7
2012-03-06 18:27:59 +11:00
# include < a s m / h w _ i r q . h >
2015-11-20 17:14:02 +08:00
# include < a s m / c p u t h r e a d s . h >
2016-03-15 01:47:38 -05:00
# include < a s m / p p c - o p c o d e . h >
2016-01-13 23:33:46 -05:00
# include < a s m / e x p o r t . h >
2018-07-05 16:25:01 +00:00
# include < a s m / f e a t u r e - f i x u p s . h >
2020-11-08 16:57:35 +00:00
# ifdef C O N F I G _ P P C _ B O O K 3 S
# include < a s m / e x c e p t i o n - 6 4 s . h >
# else
# include < a s m / e x c e p t i o n - 6 4 e . h >
# endif
2005-09-26 16:04:21 +10:00
2011-03-30 22:57:33 -03:00
/ * The p h y s i c a l m e m o r y i s l a i d o u t s u c h t h a t t h e s e c o n d a r y p r o c e s s o r
2009-06-02 21:17:38 +00:00
* spin c o d e s i t s a t 0 x00 0 0 . . . 0 x00 f f . O n s e r v e r , t h e v e c t o r s f o l l o w
* using t h e l a y o u t d e s c r i b e d i n e x c e p t i o n s - 6 4 s . S
2005-09-26 16:04:21 +10:00
* /
/ *
* Entering i n t o t h i s c o d e w e m a k e t h e f o l l o w i n g a s s u m p t i o n s :
2009-06-02 21:17:38 +00:00
*
* For p S e r i e s o r s e r v e r p r o c e s s o r s :
2005-09-26 16:04:21 +10:00
* 1 . The M M U i s o f f & o p e n f i r m w a r e i s r u n n i n g i n r e a l m o d e .
2017-10-23 18:05:07 +10:00
* 2 . The p r i m a r y C P U e n t e r s a t _ _ s t a r t .
* 3 . If t h e R T A S s u p p o r t s " q u e r y - c p u - s t o p p e d - s t a t e " , t h e n s e c o n d a r y
* CPUs w i l l e n t e r a s d i r e c t e d b y " s t a r t - c p u " R T A S c a l l , w h i c h i s
* generic_ s e c o n d a r y _ s m p _ i n i t , w i t h P I R i n r3 .
* 4 . Else t h e s e c o n d a r y C P U s w i l l e n t e r a t s e c o n d a r y _ h o l d ( 0 x60 ) a s
* directed b y t h e " s t a r t - c p u " R T S c a l l , w i t h P I R i n r3 .
2011-09-19 18:27:58 +00:00
* - or- F o r O P A L e n t r y :
2017-10-23 18:05:07 +10:00
* 1 . The M M U i s o f f , p r o c e s s o r i n H V m o d e .
* 2 . The p r i m a r y C P U e n t e r s a t 0 w i t h d e v i c e - t r e e i n r3 , O P A L b a s e
* in r8 , a n d e n t r y i n r9 f o r d e b u g g i n g p u r p o s e s .
* 3 . Secondary C P U s e n t e r a s d i r e c t e d b y O P A L _ S T A R T _ C P U c a l l , w h i c h
* is a t g e n e r i c _ s e c o n d a r y _ s m p _ i n i t , w i t h P I R i n r3 .
2005-09-26 16:04:21 +10:00
*
2009-06-02 21:17:38 +00:00
* For B o o k 3 E p r o c e s s o r s :
* 1 . The M M U i s o n r u n n i n g i n A S 0 i n a s t a t e d e f i n e d i n e P A P R
* 2 . The k e r n e l i s e n t e r e d a t _ _ s t a r t
2005-09-26 16:04:21 +10:00
* /
2016-09-28 11:31:48 +10:00
OPEN_ F I X E D _ S E C T I O N ( f i r s t _ 2 5 6 B , 0 x0 , 0 x10 0 )
USE_ F I X E D _ S E C T I O N ( f i r s t _ 2 5 6 B )
/ *
* Offsets a r e r e l a t i v e f r o m t h e s t a r t o f f i x e d s e c t i o n , a n d
* first_ 2 5 6 B s t a r t s a t 0 . O f f s e t s a r e a b i t e a s i e r t o u s e h e r e
* than t h e f i x e d s e c t i o n e n t r y m a c r o s .
* /
. = 0 x0
2005-09-26 16:04:21 +10:00
_ GLOBAL( _ _ s t a r t )
/* NOP this out unconditionally */
BEGIN_ F T R _ S E C T I O N
2013-09-23 12:04:45 +10:00
FIXUP_ E N D I A N
2014-02-04 16:04:35 +11:00
b _ _ s t a r t _ i n i t i a l i z a t i o n _ m u l t i p l a t f o r m
2005-09-26 16:04:21 +10:00
END_ F T R _ S E C T I O N ( 0 , 1 )
/* Catch branch to 0 in real mode */
trap
2014-03-11 11:54:06 +11:00
/ * Secondary p r o c e s s o r s s p i n o n t h i s v a l u e u n t i l i t b e c o m e s n o n - z e r o .
* When n o n - z e r o , i t c o n t a i n s t h e r e a l a d d r e s s o f t h e f u n c t i o n t h e c p u
* should j u m p t o .
2008-08-30 11:40:24 +10:00
* /
2013-12-28 13:01:47 -08:00
.balign 8
2005-09-26 16:04:21 +10:00
.globl __secondary_hold_spinloop
__secondary_hold_spinloop :
2017-03-09 16:42:12 +11:00
.8byte 0x0
2005-09-26 16:04:21 +10:00
/* Secondary processors write this value with their cpu # */
/* after they enter the spin loop immediately below. */
.globl __secondary_hold_acknowledge
__secondary_hold_acknowledge :
2017-03-09 16:42:12 +11:00
.8byte 0x0
2005-09-26 16:04:21 +10:00
2010-11-18 00:35:07 +00:00
# ifdef C O N F I G _ R E L O C A T A B L E
2008-10-23 18:41:09 +00:00
/ * This f l a g i s s e t t o 1 b y a l o a d e r i f t h e k e r n e l s h o u l d r u n
* at t h e l o a d e d a d d r e s s i n s t e a d o f t h e l i n k e d a d d r e s s . T h i s
2022-05-19 00:26:29 +10:00
* is u s e d b y k e x e c - t o o l s t o k e e p t h e k d u m p k e r n e l i n t h e
2008-10-23 18:41:09 +00:00
* crash_ k e r n e l r e g i o n . T h e l o a d e r i s r e s p o n s i b l e f o r
* observing t h e a l i g n m e n t r e q u i r e m e n t .
* /
2016-10-14 18:31:33 +11:00
# ifdef C O N F I G _ R E L O C A T A B L E _ T E S T
# define R U N _ A T _ L O A D _ D E F A U L T 1 / * T e s t r e l o c a t i o n , d o n o t c o p y t o 0 * /
# else
# define R U N _ A T _ L O A D _ D E F A U L T 0 x72 7 5 6 e 3 0 / * " r u n 0 " - - r e l o c a t e t o 0 b y d e f a u l t * /
# endif
2008-10-23 18:41:09 +00:00
/* Do not move this variable as kexec-tools knows about it. */
. = 0 x5 c
.globl __run_at_load
__run_at_load :
2021-12-21 16:59:02 +11:00
DEFINE_ F I X E D _ S Y M B O L ( _ _ r u n _ a t _ l o a d , f i r s t _ 2 5 6 B )
2016-10-14 18:31:33 +11:00
.long RUN_AT_LOAD_DEFAULT
2008-10-23 18:41:09 +00:00
# endif
2005-09-26 16:04:21 +10:00
. = 0 x6 0
/ *
2007-06-16 08:06:23 +10:00
* The f o l l o w i n g c o d e i s u s e d t o h o l d s e c o n d a r y p r o c e s s o r s
* in a s p i n l o o p a f t e r t h e y h a v e e n t e r e d t h e k e r n e l , b u t
2005-09-26 16:04:21 +10:00
* before t h e b u l k o f t h e k e r n e l h a s b e e n r e l o c a t e d . T h i s c o d e
* is r e l o c a t e d t o p h y s i c a l a d d r e s s 0 x60 b e f o r e p r o m _ i n i t i s r u n .
* All o f i t m u s t f i t b e l o w t h e f i r s t e x c e p t i o n v e c t o r a t 0 x10 0 .
2008-08-30 11:40:24 +10:00
* Use . g l o b l h e r e n o t _ G L O B A L b e c a u s e w e w a n t _ _ s e c o n d a r y _ h o l d
* to b e t h e a c t u a l t e x t a d d r e s s , n o t a d e s c r i p t o r .
2005-09-26 16:04:21 +10:00
* /
2008-08-30 11:40:24 +10:00
.globl __secondary_hold
__secondary_hold :
2013-09-23 12:04:45 +10:00
FIXUP_ E N D I A N
2022-09-19 19:01:33 +02:00
# ifndef C O N F I G _ P P C _ B O O K 3 E _ 6 4
2005-09-26 16:04:21 +10:00
mfmsr r24
ori r24 ,r24 ,M S R _ R I
mtmsrd r24 / * R I o n * /
2009-07-23 23:15:59 +00:00
# endif
2006-02-13 18:11:13 +11:00
/* Grab our physical cpu number */
2005-09-26 16:04:21 +10:00
mr r24 ,r3
2012-12-03 17:05:47 +00:00
/* stash r4 for book3e */
mr r25 ,r4
2005-09-26 16:04:21 +10:00
/* Tell the master cpu we're here */
/* Relocation is off & we are located at an address less */
/* than 0x100, so only need to grab low order offset. */
2021-12-21 16:59:02 +11:00
std r24 ,( A B S _ A D D R ( _ _ s e c o n d a r y _ h o l d _ a c k n o w l e d g e , f i r s t _ 2 5 6 B ) ) ( 0 )
2005-09-26 16:04:21 +10:00
sync
/* All secondary cpus wait here until told to start. */
2023-02-03 21:38:58 +10:00
100 : ld r12 ,( A B S _ A D D R ( _ _ s e c o n d a r y _ h o l d _ s p i n l o o p , f i r s t _ 2 5 6 B ) ) ( 0 )
2014-02-04 16:07:47 +11:00
cmpdi 0 ,r12 ,0
2008-08-30 11:40:24 +10:00
beq 1 0 0 b
2005-09-26 16:04:21 +10:00
2016-11-29 23:45:50 +11:00
# if d e f i n e d ( C O N F I G _ S M P ) | | d e f i n e d ( C O N F I G _ K E X E C _ C O R E )
2022-09-19 19:01:33 +02:00
# ifdef C O N F I G _ P P C _ B O O K 3 E _ 6 4
2014-02-04 16:07:47 +11:00
tovirt( r12 ,r12 )
# endif
mtctr r12
2005-09-26 16:04:21 +10:00
mr r3 ,r24
2012-12-03 17:05:47 +00:00
/ *
* it m a y b e t h e c a s e t h a t o t h e r p l a t f o r m s h a v e r4 r i g h t t o
* begin w i t h , t h i s g i v e s u s s o m e s a f e t y i n c a s e i t i s n o t
* /
2022-09-19 19:01:33 +02:00
# ifdef C O N F I G _ P P C _ B O O K 3 E _ 6 4
2012-12-03 17:05:47 +00:00
mr r4 ,r25
# else
2009-07-23 23:15:59 +00:00
li r4 ,0
2012-12-03 17:05:47 +00:00
# endif
2011-04-05 14:34:58 +10:00
/* Make sure that patched code is visible */
isync
2005-12-05 15:49:00 -06:00
bctr
2005-09-26 16:04:21 +10:00
# else
2019-08-26 11:10:23 +00:00
0 : trap
EMIT_ B U G _ E N T R Y 0 b , _ _ F I L E _ _ , _ _ L I N E _ _ , 0
2005-09-26 16:04:21 +10:00
# endif
2016-09-28 11:31:48 +10:00
CLOSE_ F I X E D _ S E C T I O N ( f i r s t _ 2 5 6 B )
2005-09-26 16:04:21 +10:00
/ *
2009-06-02 21:17:38 +00:00
* On s e r v e r , w e i n c l u d e t h e e x c e p t i o n v e c t o r s c o d e h e r e a s i t
* relies o n a b s o l u t e a d d r e s s i n g w h i c h i s o n l y p o s s i b l e w i t h i n
* this c o m p i l a t i o n u n i t
2005-11-07 11:06:55 +11:00
* /
2009-06-02 21:17:38 +00:00
# ifdef C O N F I G _ P P C _ B O O K 3 S
# include " e x c e p t i o n s - 6 4 s . S "
2016-09-28 11:31:48 +10:00
# else
OPEN_ T E X T _ S E C T I O N ( 0 x10 0 )
2008-08-30 11:40:24 +10:00
# endif
2005-11-07 11:06:55 +11:00
2016-09-28 11:31:48 +10:00
USE_ T E X T _ S E C T I O N ( )
2021-06-18 01:51:05 +10:00
# include " i n t e r r u p t _ 6 4 . S "
2022-09-19 19:01:33 +02:00
# ifdef C O N F I G _ P P C _ B O O K 3 E _ 6 4
2015-11-20 17:14:02 +08:00
/ *
* The b o o t i n g _ t h r e a d _ h w i d h o l d s t h e t h r e a d i d w e w a n t t o b o o t i n c p u
* hotplug c a s e . I t i s s e t b y c p u h o t p l u g c o d e , a n d i s i n v a l i d b y d e f a u l t .
* The t h r e a d i d i s t h e s a m e a s t h e i n i t i a l v a l u e o f S P R N _ P I R [ T H R E A D _ I D ]
* bit f i e l d .
* /
.globl booting_thread_hwid
booting_thread_hwid :
.long INVALID_THREAD_HWID
.align 3
/ *
* start a t h r e a d i n t h e s a m e c o r e
* input p a r a m e t e r s :
* r3 = t h e t h r e a d p h y s i c a l i d
* r4 = t h e e n t r y p o i n t w h e r e t h r e a d s t a r t s
* /
_ GLOBAL( b o o k 3 e _ s t a r t _ t h r e a d )
LOAD_ R E G _ I M M E D I A T E ( r5 , M S R _ K E R N E L )
2016-11-24 00:02:07 +11:00
cmpwi r3 , 0
2015-11-20 17:14:02 +08:00
beq 1 0 f
2016-11-24 00:02:07 +11:00
cmpwi r3 , 1
2015-11-20 17:14:02 +08:00
beq 1 1 f
/* If the thread id is invalid, just exit. */
b 1 3 f
10 :
2016-03-15 01:47:38 -05:00
MTTMR( T M R N _ I M S R 0 , 5 )
MTTMR( T M R N _ I N I A 0 , 4 )
2015-11-20 17:14:02 +08:00
b 1 2 f
11 :
2016-03-15 01:47:38 -05:00
MTTMR( T M R N _ I M S R 1 , 5 )
MTTMR( T M R N _ I N I A 1 , 4 )
2015-11-20 17:14:02 +08:00
12 :
isync
li r6 , 1
sld r6 , r6 , r3
mtspr S P R N _ T E N S , r6
13 :
blr
2015-11-20 17:13:59 +08:00
/ *
* stop a t h r e a d i n t h e s a m e c o r e
* input p a r a m e t e r :
* r3 = t h e t h r e a d p h y s i c a l i d
* /
_ GLOBAL( b o o k 3 e _ s t o p _ t h r e a d )
2016-11-24 00:02:07 +11:00
cmpwi r3 , 0
2015-11-20 17:13:59 +08:00
beq 1 0 f
2016-11-24 00:02:07 +11:00
cmpwi r3 , 1
2015-11-20 17:13:59 +08:00
beq 1 0 f
/* If the thread id is invalid, just exit. */
b 1 3 f
10 :
li r4 , 1
sld r4 , r4 , r3
mtspr S P R N _ T E N C , r4
13 :
blr
2011-12-08 01:20:27 -06:00
_ GLOBAL( f s l _ s e c o n d a r y _ t h r e a d _ i n i t )
2015-10-06 22:48:12 -05:00
mfspr r4 ,S P R N _ B U C S R
2011-12-08 01:20:27 -06:00
/* Enable branch prediction */
lis r3 ,B U C S R _ I N I T @h
ori r3 ,r3 ,B U C S R _ I N I T @l
mtspr S P R N _ B U C S R ,r3
isync
/ *
* Fix P I R t o m a t c h t h e l i n e a r n u m b e r i n g i n t h e d e v i c e t r e e .
*
* On e 6 5 0 0 , t h e r e s e t v a l u e o f P I R u s e s t h e l o w t h r e e b i t s f o r
* the t h r e a d w i t h i n a c o r e , a n d t h e u p p e r b i t s f o r t h e c o r e
* number. T h e r e a r e t w o t h r e a d s p e r c o r e , s o s h i f t e v e r y t h i n g
* but t h e l o w b i t r i g h t b y t w o b i t s s o t h a t t h e c p u n u m b e r i n g i s
* continuous.
2015-10-06 22:48:12 -05:00
*
* If t h e o l d v a l u e o f B U C S R i s n o n - z e r o , t h i s t h r e a d h a s r u n
* before. T h u s , w e a s s u m e w e a r e c o m i n g f r o m k e x e c o r a s i m i l a r
* scenario, a n d P I R i s a l r e a d y s e t t o t h e c o r r e c t v a l u e . T h i s
* is a b i t o f a h a c k , b u t t h e r e a r e l i m i t e d o p p o r t u n i t i e s f o r
* getting i n f o r m a t i o n i n t o t h e t h r e a d a n d t h e a l t e r n a t i v e s
* seemed l i k e t h e y ' d b e o v e r k i l l . W e c a n ' t t e l l j u s t b y l o o k i n g
* at t h e o l d P I R v a l u e w h i c h s t a t e i t ' s i n , s i n c e t h e s a m e v a l u e
* could b e v a l i d f o r o n e t h r e a d o u t o f r e s e t a n d f o r a d i f f e r e n t
* thread i n L i n u x .
2011-12-08 01:20:27 -06:00
* /
2015-10-06 22:48:12 -05:00
2011-12-08 01:20:27 -06:00
mfspr r3 , S P R N _ P I R
2015-10-06 22:48:12 -05:00
cmpwi r4 ,0
bne 1 f
2011-12-08 01:20:27 -06:00
rlwimi r3 , r3 , 3 0 , 2 , 3 0
mtspr S P R N _ P I R , r3
2015-10-06 22:48:12 -05:00
1 :
2009-07-23 23:15:59 +00:00
mr r24 ,r3
/* turn on 64-bit mode */
2014-02-04 16:04:35 +11:00
bl e n a b l e _ 6 4 b _ m o d e
2009-07-23 23:15:59 +00:00
/* get a valid TOC pointer, wherever we're mapped at */
2014-02-04 16:04:35 +11:00
bl r e l a t i v e _ t o c
2012-11-26 17:41:08 +00:00
tovirt( r2 ,r2 )
2009-07-23 23:15:59 +00:00
/* Book3E initialization */
mr r3 ,r24
2014-02-04 16:04:35 +11:00
bl b o o k 3 e _ s e c o n d a r y _ t h r e a d _ i n i t
2009-07-23 23:15:59 +00:00
b g e n e r i c _ s e c o n d a r y _ c o m m o n _ i n i t
2005-09-26 16:04:21 +10:00
2022-09-19 19:01:33 +02:00
# endif / * C O N F I G _ P P C _ B O O K 3 E _ 6 4 * /
2020-08-19 11:57:04 +10:00
2005-09-26 16:04:21 +10:00
/ *
2006-08-11 00:07:08 -05:00
* On p S e r i e s a n d m o s t o t h e r p l a t f o r m s , s e c o n d a r y p r o c e s s o r s s p i n
* in t h e f o l l o w i n g c o d e .
2005-09-26 16:04:21 +10:00
* At e n t r y , r3 = t h i s p r o c e s s o r ' s n u m b e r ( p h y s i c a l c p u i d )
2009-07-23 23:15:59 +00:00
*
* On B o o k 3 E , r4 = 1 t o i n d i c a t e t h a t t h e i n i t i a l T L B e n t r y f o r
* this c o r e a l r e a d y e x i s t s ( s e t u p v i a s o m e o t h e r m e c h a n i s m s u c h
* as S C O M b e f o r e e n t r y ) .
2005-09-26 16:04:21 +10:00
* /
2006-08-11 00:07:08 -05:00
_ GLOBAL( g e n e r i c _ s e c o n d a r y _ s m p _ i n i t )
2013-09-23 12:04:45 +10:00
FIXUP_ E N D I A N
2005-09-26 16:04:21 +10:00
mr r24 ,r3
2009-07-23 23:15:59 +00:00
mr r25 ,r4
2005-09-26 16:04:21 +10:00
/* turn on 64-bit mode */
2014-02-04 16:04:35 +11:00
bl e n a b l e _ 6 4 b _ m o d e
2005-09-26 16:04:21 +10:00
2009-07-23 23:15:59 +00:00
/* get a valid TOC pointer, wherever we're mapped at */
2014-02-04 16:04:35 +11:00
bl r e l a t i v e _ t o c
2012-11-26 17:41:08 +00:00
tovirt( r2 ,r2 )
2008-08-30 11:41:12 +10:00
2022-09-19 19:01:33 +02:00
# ifdef C O N F I G _ P P C _ B O O K 3 E _ 6 4
2009-07-23 23:15:59 +00:00
/* Book3E initialization */
mr r3 ,r24
mr r4 ,r25
2014-02-04 16:04:35 +11:00
bl b o o k 3 e _ s e c o n d a r y _ c o r e _ i n i t
2015-11-20 17:14:02 +08:00
/ *
* After c o m m o n c o r e i n i t h a s f i n i s h e d , c h e c k i f t h e c u r r e n t t h r e a d i s t h e
* one w e w a n t e d t o b o o t . I f n o t , s t a r t t h e s p e c i f i e d t h r e a d a n d s t o p t h e
* current t h r e a d .
* /
LOAD_ R E G _ A D D R ( r4 , b o o t i n g _ t h r e a d _ h w i d )
lwz r3 , 0 ( r4 )
li r5 , I N V A L I D _ T H R E A D _ H W I D
cmpw r3 , r5
beq 2 0 f
/ *
* The v a l u e o f b o o t i n g _ t h r e a d _ h w i d h a s b e e n s t o r e d i n r3 ,
* so m a k e i t i n v a l i d .
* /
stw r5 , 0 ( r4 )
/ *
* Get t h e c u r r e n t t h r e a d i d a n d c h e c k i f i t i s t h e o n e w e w a n t e d .
* If n o t , s t a r t t h e o n e s p e c i f i e d i n b o o t i n g _ t h r e a d _ h w i d a n d s t o p
* the c u r r e n t t h r e a d .
* /
mfspr r8 , S P R N _ T I R
cmpw r3 , r8
beq 2 0 f
/* start the specified thread */
LOAD_ R E G _ A D D R ( r5 , f s l _ s e c o n d a r y _ t h r e a d _ i n i t )
ld r4 , 0 ( r5 )
bl b o o k 3 e _ s t a r t _ t h r e a d
/* stop the current thread */
mr r3 , r8
bl b o o k 3 e _ s t o p _ t h r e a d
10 :
b 1 0 b
20 :
2009-07-23 23:15:59 +00:00
# endif
generic_secondary_common_init :
2005-09-26 16:04:21 +10:00
/ * Set u p a p a c a v a l u e f o r t h i s p r o c e s s o r . S i n c e w e h a v e t h e
* physical c p u i d i n r24 , w e n e e d t o s e a r c h t h e p a c a s t o f i n d
* which l o g i c a l i d m a p s t o o u r p h y s i c a l o n e .
* /
2011-05-10 19:28:37 +00:00
# ifndef C O N F I G _ S M P
2014-02-04 16:04:35 +11:00
b k e x e c _ w a i t / * w a i t f o r n e x t k e r n e l i f ! S M P * /
2011-05-10 19:28:37 +00:00
# else
2018-02-14 01:08:12 +10:00
LOAD_ R E G _ A D D R ( r8 , p a c a _ p t r s ) / * L o a d p a c a _ p t r s p o i n t e * /
ld r8 ,0 ( r8 ) / * G e t b a s e v a d d r o f a r r a y * /
powerpc/64: don't refer nr_cpu_ids in asm code when it's undefined
generic_secondary_common_init() calls LOAD_REG_ADDR(r7, nr_cpu_ids)
conditionally on CONFIG_SMP. However, if 'NR_CPUS == 1', kernel doesn't
use the nr_cpu_ids, and in C code, it's just:
#if NR_CPUS == 1
#define nr_cpu_ids
...
This series makes declaration of nr_cpu_ids conditional on NR_CPUS == 1,
and that reveals the issue, because compiler can't link the
LOAD_REG_ADDR(r7, nr_cpu_ids) against nonexisting symbol.
Current code looks unsafe for those who build kernel with CONFIG_SMP=y and
NR_CPUS == 1. This is weird configuration, but not disallowed.
Fix the linker error by replacing LOAD_REG_ADDR() with LOAD_REG_IMMEDIATE()
conditionally on NR_CPUS == 1.
As the following patch adds CONFIG_FORCE_NR_CPUS option that has the
similar effect on nr_cpu_ids, make the generic_secondary_common_init()
conditional on it too.
Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Yury Norov <yury.norov@gmail.com>
2022-09-20 06:13:12 -07:00
# if ( N R _ C P U S = = 1 ) | | d e f i n e d ( C O N F I G _ F O R C E _ N R _ C P U S )
LOAD_ R E G _ I M M E D I A T E ( r7 , N R _ C P U S )
# else
2011-05-10 19:28:37 +00:00
LOAD_ R E G _ A D D R ( r7 , n r _ c p u _ i d s ) / * L o a d n r _ c p u _ i d s a d d r e s s * /
lwz r7 ,0 ( r7 ) / * a l s o t h e m a x p a c a a l l o c a t e d * /
powerpc/64: don't refer nr_cpu_ids in asm code when it's undefined
generic_secondary_common_init() calls LOAD_REG_ADDR(r7, nr_cpu_ids)
conditionally on CONFIG_SMP. However, if 'NR_CPUS == 1', kernel doesn't
use the nr_cpu_ids, and in C code, it's just:
#if NR_CPUS == 1
#define nr_cpu_ids
...
This series makes declaration of nr_cpu_ids conditional on NR_CPUS == 1,
and that reveals the issue, because compiler can't link the
LOAD_REG_ADDR(r7, nr_cpu_ids) against nonexisting symbol.
Current code looks unsafe for those who build kernel with CONFIG_SMP=y and
NR_CPUS == 1. This is weird configuration, but not disallowed.
Fix the linker error by replacing LOAD_REG_ADDR() with LOAD_REG_IMMEDIATE()
conditionally on NR_CPUS == 1.
As the following patch adds CONFIG_FORCE_NR_CPUS option that has the
similar effect on nr_cpu_ids, make the generic_secondary_common_init()
conditional on it too.
Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Yury Norov <yury.norov@gmail.com>
2022-09-20 06:13:12 -07:00
# endif
2005-09-26 16:04:21 +10:00
li r5 ,0 / * l o g i c a l c p u i d * /
2018-02-14 01:08:12 +10:00
1 :
sldi r9 ,r5 ,3 / * g e t p a c a _ p t r s [ ] i n d e x f r o m c p u i d * /
ldx r13 ,r9 ,r8 / * r13 = p a c a _ p t r s [ c p u i d ] * /
lhz r6 ,P A C A H W C P U I D ( r13 ) / * L o a d H W p r o c i d f r o m p a c a * /
2005-09-26 16:04:21 +10:00
cmpw r6 ,r24 / * C o m p a r e t o o u r i d * /
beq 2 f
addi r5 ,r5 ,1
2011-05-10 19:28:37 +00:00
cmpw r5 ,r7 / * C h e c k i f m o r e p a c a s e x i s t * /
2005-09-26 16:04:21 +10:00
blt 1 b
mr r3 ,r24 / * n o t f o u n d , c o p y p h y s t o r3 * /
2014-02-04 16:04:35 +11:00
b k e x e c _ w a i t / * n e x t k e r n e l m i g h t d o b e t t e r * /
2005-09-26 16:04:21 +10:00
2011-01-20 17:50:21 +11:00
2 : SET_ P A C A ( r13 )
2022-09-19 19:01:33 +02:00
# ifdef C O N F I G _ P P C _ B O O K 3 E _ 6 4
2009-07-23 23:15:59 +00:00
addi r12 ,r13 ,P A C A _ E X T L B / * a n d T L B e x c f r a m e i n a n o t h e r * /
mtspr S P R N _ S P R G _ T L B _ E X F R A M E ,r12
# endif
2005-09-26 16:04:21 +10:00
/* From now on, r24 is expected to be logical cpuid */
mr r24 ,r5
2008-07-12 09:00:26 +10:00
2020-10-14 18:28:36 +11:00
/* Create a temp kernel stack for use before relocation is on. */
ld r1 ,P A C A E M E R G S P ( r13 )
2022-11-27 22:49:40 +10:00
subi r1 ,r1 ,S T A C K _ F R A M E _ M I N _ S I Z E
2020-10-14 18:28:36 +11:00
2006-08-11 00:07:08 -05:00
/* See if we need to call a cpu state restore handler */
2008-08-30 11:41:12 +10:00
LOAD_ R E G _ A D D R ( r23 , c u r _ c p u _ s p e c )
2006-08-11 00:07:08 -05:00
ld r23 ,0 ( r23 )
2014-03-11 11:54:06 +11:00
ld r12 ,C P U _ S P E C _ R E S T O R E ( r23 )
cmpdi 0 ,r12 ,0
2011-03-16 14:54:35 +11:00
beq 3 f
2022-05-09 07:36:07 +02:00
# ifdef C O N F I G _ P P C 6 4 _ E L F _ A B I _ V 1
2014-03-11 11:54:06 +11:00
ld r12 ,0 ( r12 )
# endif
2014-02-04 16:07:47 +11:00
mtctr r12
2006-08-11 00:07:08 -05:00
bctrl
2011-05-25 18:09:12 +00:00
3 : LOAD_ R E G _ A D D R ( r3 , s p i n n i n g _ s e c o n d a r i e s ) / * D e c r e m e n t s p i n n i n g _ s e c o n d a r i e s * /
2011-03-16 14:54:35 +11:00
lwarx r4 ,0 ,r3
subi r4 ,r4 ,1
stwcx. r4 ,0 ,r3
bne 3 b
isync
4 : HMT_ L O W
2011-02-01 12:13:09 +11:00
lbz r23 ,P A C A P R O C S T A R T ( r13 ) / * T e s t i f t h i s p r o c e s s o r s h o u l d * /
/* start. */
cmpwi 0 ,r23 ,0
2011-03-16 14:54:35 +11:00
beq 4 b / * L o o p u n t i l t o l d t o g o * /
2011-02-01 12:13:09 +11:00
sync / * o r d e r p a c a . r u n a n d c u r _ c p u _ s p e c * /
2011-03-16 14:54:35 +11:00
isync / * I n c a s e c o d e p a t c h i n g h a p p e n e d * /
2011-02-01 12:13:09 +11:00
2006-11-27 14:59:50 +11:00
b _ _ s e c o n d a r y _ s t a r t
2011-05-10 19:28:37 +00:00
# endif / * S M P * /
2005-09-26 16:04:21 +10:00
2008-08-30 11:41:12 +10:00
/ *
* Turn t h e M M U o f f .
* Assumes w e ' r e m a p p e d E A = = R A i f t h e M M U i s o n .
* /
2009-07-23 23:15:59 +00:00
# ifdef C O N F I G _ P P C _ B O O K 3 S
2022-11-14 23:27:41 +05:30
SYM_ F U N C _ S T A R T _ L O C A L ( _ _ m m u _ o f f )
2005-09-26 16:04:21 +10:00
mfmsr r3
andi. r0 ,r3 ,M S R _ I R | M S R _ D R
beqlr
2008-08-30 11:41:12 +10:00
mflr r4
2005-09-26 16:04:21 +10:00
andc r3 ,r3 ,r0
mtspr S P R N _ S R R 0 ,r4
mtspr S P R N _ S R R 1 ,r3
sync
rfid
b . / * p r e v e n t s p e c u l a t i v e e x e c u t i o n * /
2022-11-14 23:27:41 +05:30
SYM_ F U N C _ E N D ( _ _ m m u _ o f f )
2005-09-26 16:04:21 +10:00
2023-02-17 10:02:26 +05:30
SYM_ F U N C _ S T A R T _ L O C A L ( s t a r t _ i n i t i a l i z a t i o n _ b o o k 3 s )
2023-02-03 21:38:57 +10:00
mflr r25
/* Setup some critical 970 SPRs before switching MMU off */
mfspr r0 ,S P R N _ P V R
srwi r0 ,r0 ,1 6
cmpwi r0 ,0 x39 / * 9 7 0 * /
beq 1 f
cmpwi r0 ,0 x3 c / * 9 7 0 F X * /
beq 1 f
cmpwi r0 ,0 x44 / * 9 7 0 M P * /
beq 1 f
cmpwi r0 ,0 x45 / * 9 7 0 G X * /
bne 2 f
1 : bl _ _ c p u _ p r e i n i t _ p p c97 0
2 :
/* Switch off MMU if not already off */
bl _ _ m m u _ o f f
mtlr r25
blr
2023-02-17 10:02:26 +05:30
SYM_ F U N C _ E N D ( s t a r t _ i n i t i a l i z a t i o n _ b o o k 3 s )
2023-02-03 21:38:57 +10:00
# endif
2005-09-26 16:04:21 +10:00
/ *
* Here i s o u r m a i n k e r n e l e n t r y p o i n t . W e s u p p o r t c u r r e n t l y 2 k i n d o f e n t r i e s
* depending o n t h e v a l u e o f r5 .
*
* r5 ! = N U L L - > O F e n t r y , w e g o t o p r o m _ i n i t , " l e g a c y " p a r a m e t e r c o n t e n t
* in r3 . . . r7
*
* r5 = = N U L L - > k e x e c s t y l e e n t r y . r3 i s a p h y s i c a l p o i n t e r t o t h e
* DT b l o c k , r4 i s a p h y s i c a l p o i n t e r t o t h e k e r n e l i t s e l f
*
* /
2014-02-04 16:06:11 +11:00
__start_initialization_multiplatform :
2008-08-30 11:41:12 +10:00
/* Make sure we are running in 64 bits mode */
2014-02-04 16:04:35 +11:00
bl e n a b l e _ 6 4 b _ m o d e
2008-08-30 11:41:12 +10:00
2022-09-26 15:56:20 +10:00
/* Zero r13 (paca) so early program check / mce don't use it */
li r13 ,0
2005-09-26 16:04:21 +10:00
/ *
* Are w e b o o t e d f r o m a P R O M O f - t y p e c l i e n t - i n t e r f a c e ?
* /
cmpldi c r0 ,r5 ,0
2007-07-31 16:44:13 +10:00
beq 1 f
2014-02-04 16:04:35 +11:00
b _ _ b o o t _ f r o m _ p r o m / * y e s - > p r o m * /
2007-07-31 16:44:13 +10:00
1 :
2005-09-26 16:04:21 +10:00
/* Save parameters */
mr r31 ,r3
mr r30 ,r4
2011-09-19 17:44:59 +00:00
# ifdef C O N F I G _ P P C _ E A R L Y _ D E B U G _ O P A L
/* Save OPAL entry */
mr r28 ,r8
mr r29 ,r9
# endif
2005-09-26 16:04:21 +10:00
2023-04-08 12:17:47 +10:00
/* Get TOC pointer (current runtime address) */
bl r e l a t i v e _ t o c
2022-09-19 19:01:33 +02:00
# ifdef C O N F I G _ P P C _ B O O K 3 E _ 6 4
2014-02-04 16:04:35 +11:00
bl s t a r t _ i n i t i a l i z a t i o n _ b o o k 3 e
2009-07-23 23:15:59 +00:00
# else
2023-02-03 21:38:57 +10:00
bl s t a r t _ i n i t i a l i z a t i o n _ b o o k 3 s
2022-09-19 19:01:33 +02:00
# endif / * C O N F I G _ P P C _ B O O K 3 E _ 6 4 * /
2023-04-08 12:17:47 +10:00
/* Get TOC pointer */
bl r e l a t i v e _ t o c
/* find out where we are now */
bcl 2 0 ,3 1 ,$ + 4
0 : mflr r26 / * r26 = r u n t i m e a d d r h e r e * /
addis r26 ,r26 ,( _ s t e x t - 0 b ) @ha
addi r26 ,r26 ,( _ s t e x t - 0 b ) @l /* current runtime base addr */
2023-02-03 21:38:57 +10:00
b _ _ a f t e r _ p r o m _ s t a r t
2005-09-26 16:04:21 +10:00
powerpc/64s: Fix section mismatch warnings from boot code
We currently have two section mismatch warnings:
The function __boot_from_prom() references
the function __init prom_init().
The function start_here_common() references
the function __init start_kernel().
The warnings are correct, we do have branches from non-init code into
init code, which is freed after boot. But we don't expect to ever
execute any of that early boot code after boot, if we did that would
be a bug. In particular calling into OF after boot would be fatal
because OF is no longer resident.
So for now fix the warnings by marking the relevant functions as
__REF, which puts them in the ".ref.text" section.
This causes some reordering of the functions in the final link:
@@ -217,10 +217,9 @@
c00000000000b088 t generic_secondary_common_init
c00000000000b124 t __mmu_off
c00000000000b14c t __start_initialization_multiplatform
-c00000000000b1ac t __boot_from_prom
-c00000000000b1ec t __after_prom_start
-c00000000000b260 t p_end
-c00000000000b27c T copy_and_flush
+c00000000000b1ac t __after_prom_start
+c00000000000b220 t p_end
+c00000000000b23c T copy_and_flush
c00000000000b300 T __secondary_start
c00000000000b300 t copy_to_here
c00000000000b344 t start_secondary_prolog
@@ -228,8 +227,9 @@
c00000000000b36c t enable_64b_mode
c00000000000b388 T relative_toc
c00000000000b3a8 t p_toc
-c00000000000b3b0 t start_here_common
-c00000000000b3d0 t start_here_multiplatform
+c00000000000b3b0 t __boot_from_prom
+c00000000000b3f0 t start_here_multiplatform
+c00000000000b480 t start_here_common
c00000000000b880 T system_call_common
c00000000000b974 t system_call
c00000000000b9dc t system_call_exit
In particular __boot_from_prom moves after copy_to_here, which means
it's not copied to zero in the first stage of copy of the kernel to
zero.
But that's OK, because we only call __boot_from_prom before we do the
copy, so it makes no difference when it's copied. The call sequence
is:
__start
-> __start_initialization_multiplatform
-> __boot_from_prom
-> __start
-> __start_initialization_multiplatform
-> __after_prom_start
-> copy_and_flush
-> copy_and_flush (relocated to 0)
-> start_here_multiplatform
-> early_setup
Reported-by: Mauricio Faria de Oliveira <mauricfo@linux.ibm.com>
Reported-by: Roman Bolshakov <r.bolshakov@yadro.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20200225031328.14676-1-mpe@ellerman.id.au
2020-02-25 14:13:28 +11:00
_ _ REF
2014-02-04 16:06:11 +11:00
__boot_from_prom :
2009-03-10 17:53:27 +00:00
# ifdef C O N F I G _ P P C _ O F _ B O O T _ T R A M P O L I N E
2023-04-08 12:17:47 +10:00
/* Get TOC pointer */
bl r e l a t i v e _ t o c
/* find out where we are now */
bcl 2 0 ,3 1 ,$ + 4
0 : mflr r26 / * r26 = r u n t i m e a d d r h e r e * /
addis r26 ,r26 ,( _ s t e x t - 0 b ) @ha
addi r26 ,r26 ,( _ s t e x t - 0 b ) @l /* current runtime base addr */
2005-09-26 16:04:21 +10:00
/* Save parameters */
mr r31 ,r3
mr r30 ,r4
mr r29 ,r5
mr r28 ,r6
mr r27 ,r7
[PATCH] correct the comment about stackpointer alignment in __boot_from_prom
The address of variable val in prom_init_stdout is passed to prom_getprop.
prom_getprop casts the pointer to u32 and passes it to call_prom in the hope
that OpenFirmware stores something there.
But the pointer is truncated in the lower bits and the expected value is
stored somewhere else.
In my testing I had a stackpointer of 0x0023e6b4. val was at offset 120,
wich has address 0x0023e72c. But the value passed to OF was 0x0023e728.
c00000000040b710: 3b 01 00 78 addi r24,r1,120
...
c00000000040b754: 57 08 00 38 rlwinm r8,r24,0,0,28
...
c00000000040b784: 80 01 00 78 lwz r0,120(r1)
...
c00000000040b798: 90 1b 00 0c stw r0,12(r27)
...
The stackpointer came from 32bit code.
The chain was yaboot -> zImage -> vmlinux
PowerMac OpenFirmware does appearently not handle the ELF sections
correctly. If yaboot was compiled in
/usr/src/packages/BUILD/lilo-10.1.1/yaboot, then the stackpointer is
unaligned. But the stackpointer is correct if yaboot is compiled in
/tmp/yaboot.
This bug triggered since 2.6.15, now prom_getprop is an inline
function. gcc clears the lower bits, instead of just clearing the
upper 32 bits.
Signed-off-by: Olaf Hering <olh@suse.de>
Signed-off-by: Paul Mackerras <paulus@samba.org>
2006-03-23 21:50:59 +01:00
/ *
* Align t h e s t a c k t o 1 6 - b y t e b o u n d a r y
* Depending o n t h e s i z e a n d l a y o u t o f t h e E L F s e c t i o n s i n t h e i n i t i a l
2008-08-30 11:41:12 +10:00
* boot b i n a r y , t h e s t a c k p o i n t e r m a y b e u n a l i g n e d o n P o w e r M a c
[PATCH] correct the comment about stackpointer alignment in __boot_from_prom
The address of variable val in prom_init_stdout is passed to prom_getprop.
prom_getprop casts the pointer to u32 and passes it to call_prom in the hope
that OpenFirmware stores something there.
But the pointer is truncated in the lower bits and the expected value is
stored somewhere else.
In my testing I had a stackpointer of 0x0023e6b4. val was at offset 120,
wich has address 0x0023e72c. But the value passed to OF was 0x0023e728.
c00000000040b710: 3b 01 00 78 addi r24,r1,120
...
c00000000040b754: 57 08 00 38 rlwinm r8,r24,0,0,28
...
c00000000040b784: 80 01 00 78 lwz r0,120(r1)
...
c00000000040b798: 90 1b 00 0c stw r0,12(r27)
...
The stackpointer came from 32bit code.
The chain was yaboot -> zImage -> vmlinux
PowerMac OpenFirmware does appearently not handle the ELF sections
correctly. If yaboot was compiled in
/usr/src/packages/BUILD/lilo-10.1.1/yaboot, then the stackpointer is
unaligned. But the stackpointer is correct if yaboot is compiled in
/tmp/yaboot.
This bug triggered since 2.6.15, now prom_getprop is an inline
function. gcc clears the lower bits, instead of just clearing the
upper 32 bits.
Signed-off-by: Olaf Hering <olh@suse.de>
Signed-off-by: Paul Mackerras <paulus@samba.org>
2006-03-23 21:50:59 +01:00
* /
2006-03-04 15:00:45 -08:00
rldicr r1 ,r1 ,0 ,5 9
powerpc: Make the 64-bit kernel as a position-independent executable
This implements CONFIG_RELOCATABLE for 64-bit by making the kernel as
a position-independent executable (PIE) when it is set. This involves
processing the dynamic relocations in the image in the early stages of
booting, even if the kernel is being run at the address it is linked at,
since the linker does not necessarily fill in words in the image for
which there are dynamic relocations. (In fact the linker does fill in
such words for 64-bit executables, though not for 32-bit executables,
so in principle we could avoid calling relocate() entirely when we're
running a 64-bit kernel at the linked address.)
The dynamic relocations are processed by a new function relocate(addr),
where the addr parameter is the virtual address where the image will be
run. In fact we call it twice; once before calling prom_init, and again
when starting the main kernel. This means that reloc_offset() returns
0 in prom_init (since it has been relocated to the address it is running
at), which necessitated a few adjustments.
This also changes __va and __pa to use an equivalent definition that is
simpler. With the relocatable kernel, PAGE_OFFSET and MEMORY_START are
constants (for 64-bit) whereas PHYSICAL_START is a variable (and
KERNELBASE ideally should be too, but isn't yet).
With this, relocatable kernels still copy themselves down to physical
address 0 and run there.
Signed-off-by: Paul Mackerras <paulus@samba.org>
2008-08-30 11:43:47 +10:00
# ifdef C O N F I G _ R E L O C A T A B L E
/* Relocate code for where we are now */
mr r3 ,r26
2014-02-04 16:04:35 +11:00
bl r e l o c a t e
powerpc: Make the 64-bit kernel as a position-independent executable
This implements CONFIG_RELOCATABLE for 64-bit by making the kernel as
a position-independent executable (PIE) when it is set. This involves
processing the dynamic relocations in the image in the early stages of
booting, even if the kernel is being run at the address it is linked at,
since the linker does not necessarily fill in words in the image for
which there are dynamic relocations. (In fact the linker does fill in
such words for 64-bit executables, though not for 32-bit executables,
so in principle we could avoid calling relocate() entirely when we're
running a 64-bit kernel at the linked address.)
The dynamic relocations are processed by a new function relocate(addr),
where the addr parameter is the virtual address where the image will be
run. In fact we call it twice; once before calling prom_init, and again
when starting the main kernel. This means that reloc_offset() returns
0 in prom_init (since it has been relocated to the address it is running
at), which necessitated a few adjustments.
This also changes __va and __pa to use an equivalent definition that is
simpler. With the relocatable kernel, PAGE_OFFSET and MEMORY_START are
constants (for 64-bit) whereas PHYSICAL_START is a variable (and
KERNELBASE ideally should be too, but isn't yet).
With this, relocatable kernels still copy themselves down to physical
address 0 and run there.
Signed-off-by: Paul Mackerras <paulus@samba.org>
2008-08-30 11:43:47 +10:00
# endif
2005-09-26 16:04:21 +10:00
/* Restore parameters */
mr r3 ,r31
mr r4 ,r30
mr r5 ,r29
mr r6 ,r28
mr r7 ,r27
/* Do all of the interaction with OF client interface */
powerpc: Make the 64-bit kernel as a position-independent executable
This implements CONFIG_RELOCATABLE for 64-bit by making the kernel as
a position-independent executable (PIE) when it is set. This involves
processing the dynamic relocations in the image in the early stages of
booting, even if the kernel is being run at the address it is linked at,
since the linker does not necessarily fill in words in the image for
which there are dynamic relocations. (In fact the linker does fill in
such words for 64-bit executables, though not for 32-bit executables,
so in principle we could avoid calling relocate() entirely when we're
running a 64-bit kernel at the linked address.)
The dynamic relocations are processed by a new function relocate(addr),
where the addr parameter is the virtual address where the image will be
run. In fact we call it twice; once before calling prom_init, and again
when starting the main kernel. This means that reloc_offset() returns
0 in prom_init (since it has been relocated to the address it is running
at), which necessitated a few adjustments.
This also changes __va and __pa to use an equivalent definition that is
simpler. With the relocatable kernel, PAGE_OFFSET and MEMORY_START are
constants (for 64-bit) whereas PHYSICAL_START is a variable (and
KERNELBASE ideally should be too, but isn't yet).
With this, relocatable kernels still copy themselves down to physical
address 0 and run there.
Signed-off-by: Paul Mackerras <paulus@samba.org>
2008-08-30 11:43:47 +10:00
mr r8 ,r26
2014-02-04 16:04:35 +11:00
bl p r o m _ i n i t
2009-03-10 17:53:27 +00:00
# endif / * #C O N F I G _ P P C _ O F _ B O O T _ T R A M P O L I N E * /
/ * We n e v e r r e t u r n . W e a l s o h i t t h a t t r a p i f t r y i n g t o b o o t
* from O F w h i l e C O N F I G _ P P C _ O F _ B O O T _ T R A M P O L I N E i s n ' t s e l e c t e d * /
2005-09-26 16:04:21 +10:00
trap
powerpc/64s: Fix section mismatch warnings from boot code
We currently have two section mismatch warnings:
The function __boot_from_prom() references
the function __init prom_init().
The function start_here_common() references
the function __init start_kernel().
The warnings are correct, we do have branches from non-init code into
init code, which is freed after boot. But we don't expect to ever
execute any of that early boot code after boot, if we did that would
be a bug. In particular calling into OF after boot would be fatal
because OF is no longer resident.
So for now fix the warnings by marking the relevant functions as
__REF, which puts them in the ".ref.text" section.
This causes some reordering of the functions in the final link:
@@ -217,10 +217,9 @@
c00000000000b088 t generic_secondary_common_init
c00000000000b124 t __mmu_off
c00000000000b14c t __start_initialization_multiplatform
-c00000000000b1ac t __boot_from_prom
-c00000000000b1ec t __after_prom_start
-c00000000000b260 t p_end
-c00000000000b27c T copy_and_flush
+c00000000000b1ac t __after_prom_start
+c00000000000b220 t p_end
+c00000000000b23c T copy_and_flush
c00000000000b300 T __secondary_start
c00000000000b300 t copy_to_here
c00000000000b344 t start_secondary_prolog
@@ -228,8 +227,9 @@
c00000000000b36c t enable_64b_mode
c00000000000b388 T relative_toc
c00000000000b3a8 t p_toc
-c00000000000b3b0 t start_here_common
-c00000000000b3d0 t start_here_multiplatform
+c00000000000b3b0 t __boot_from_prom
+c00000000000b3f0 t start_here_multiplatform
+c00000000000b480 t start_here_common
c00000000000b880 T system_call_common
c00000000000b974 t system_call
c00000000000b9dc t system_call_exit
In particular __boot_from_prom moves after copy_to_here, which means
it's not copied to zero in the first stage of copy of the kernel to
zero.
But that's OK, because we only call __boot_from_prom before we do the
copy, so it makes no difference when it's copied. The call sequence
is:
__start
-> __start_initialization_multiplatform
-> __boot_from_prom
-> __start
-> __start_initialization_multiplatform
-> __after_prom_start
-> copy_and_flush
-> copy_and_flush (relocated to 0)
-> start_here_multiplatform
-> early_setup
Reported-by: Mauricio Faria de Oliveira <mauricfo@linux.ibm.com>
Reported-by: Roman Bolshakov <r.bolshakov@yadro.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20200225031328.14676-1-mpe@ellerman.id.au
2020-02-25 14:13:28 +11:00
.previous
2005-09-26 16:04:21 +10:00
2014-02-04 16:06:11 +11:00
__after_prom_start :
powerpc: Make the 64-bit kernel as a position-independent executable
This implements CONFIG_RELOCATABLE for 64-bit by making the kernel as
a position-independent executable (PIE) when it is set. This involves
processing the dynamic relocations in the image in the early stages of
booting, even if the kernel is being run at the address it is linked at,
since the linker does not necessarily fill in words in the image for
which there are dynamic relocations. (In fact the linker does fill in
such words for 64-bit executables, though not for 32-bit executables,
so in principle we could avoid calling relocate() entirely when we're
running a 64-bit kernel at the linked address.)
The dynamic relocations are processed by a new function relocate(addr),
where the addr parameter is the virtual address where the image will be
run. In fact we call it twice; once before calling prom_init, and again
when starting the main kernel. This means that reloc_offset() returns
0 in prom_init (since it has been relocated to the address it is running
at), which necessitated a few adjustments.
This also changes __va and __pa to use an equivalent definition that is
simpler. With the relocatable kernel, PAGE_OFFSET and MEMORY_START are
constants (for 64-bit) whereas PHYSICAL_START is a variable (and
KERNELBASE ideally should be too, but isn't yet).
With this, relocatable kernels still copy themselves down to physical
address 0 and run there.
Signed-off-by: Paul Mackerras <paulus@samba.org>
2008-08-30 11:43:47 +10:00
# ifdef C O N F I G _ R E L O C A T A B L E
/* process relocations for the final address of the kernel */
lis r25 ,P A G E _ O F F S E T @highest /* compute virtual base of kernel */
sldi r25 ,r25 ,3 2
2022-09-19 19:01:33 +02:00
# if d e f i n e d ( C O N F I G _ P P C _ B O O K 3 E _ 6 4 )
2015-10-06 22:48:15 -05:00
tovirt( r26 ,r26 ) / * o n b o o k e , w e a l r e a d y r u n a t P A G E _ O F F S E T * /
# endif
2016-09-28 11:31:48 +10:00
lwz r7 ,( F I X E D _ S Y M B O L _ A B S _ A D D R ( _ _ r u n _ a t _ l o a d ) ) ( r26 )
2022-09-19 19:01:33 +02:00
# if d e f i n e d ( C O N F I G _ P P C _ B O O K 3 E _ 6 4 )
2015-10-06 22:48:15 -05:00
tophys( r26 ,r26 )
# endif
2010-11-18 00:35:07 +00:00
cmplwi c r0 ,r7 ,1 / * f l a g g e d t o s t a y w h e r e w e a r e ? * /
2008-10-21 17:38:10 +00:00
bne 1 f
add r25 ,r25 ,r26
1 : mr r3 ,r25
2014-02-04 16:04:35 +11:00
bl r e l o c a t e
2022-09-19 19:01:33 +02:00
# if d e f i n e d ( C O N F I G _ P P C _ B O O K 3 E _ 6 4 )
2015-10-06 22:48:15 -05:00
/* IVPR needs to be set after relocation. */
bl i n i t _ c o r e _ b o o k 3 e
# endif
powerpc: Make the 64-bit kernel as a position-independent executable
This implements CONFIG_RELOCATABLE for 64-bit by making the kernel as
a position-independent executable (PIE) when it is set. This involves
processing the dynamic relocations in the image in the early stages of
booting, even if the kernel is being run at the address it is linked at,
since the linker does not necessarily fill in words in the image for
which there are dynamic relocations. (In fact the linker does fill in
such words for 64-bit executables, though not for 32-bit executables,
so in principle we could avoid calling relocate() entirely when we're
running a 64-bit kernel at the linked address.)
The dynamic relocations are processed by a new function relocate(addr),
where the addr parameter is the virtual address where the image will be
run. In fact we call it twice; once before calling prom_init, and again
when starting the main kernel. This means that reloc_offset() returns
0 in prom_init (since it has been relocated to the address it is running
at), which necessitated a few adjustments.
This also changes __va and __pa to use an equivalent definition that is
simpler. With the relocatable kernel, PAGE_OFFSET and MEMORY_START are
constants (for 64-bit) whereas PHYSICAL_START is a variable (and
KERNELBASE ideally should be too, but isn't yet).
With this, relocatable kernels still copy themselves down to physical
address 0 and run there.
Signed-off-by: Paul Mackerras <paulus@samba.org>
2008-08-30 11:43:47 +10:00
# endif
2005-09-26 16:04:21 +10:00
/ *
2008-08-30 11:41:12 +10:00
* We n e e d t o r u n w i t h _ s t e x t a t p h y s i c a l a d d r e s s P H Y S I C A L _ S T A R T .
2005-09-26 16:04:21 +10:00
* This w i l l l e a v e s o m e c o d e i n t h e f i r s t 2 5 6 B o f
* real m e m o r y , w h i c h a r e r e s e r v e d f o r s o f t w a r e u s e .
*
* Note : This p r o c e s s o v e r w r i t e s t h e O F e x c e p t i o n v e c t o r s .
* /
powerpc: Make the 64-bit kernel as a position-independent executable
This implements CONFIG_RELOCATABLE for 64-bit by making the kernel as
a position-independent executable (PIE) when it is set. This involves
processing the dynamic relocations in the image in the early stages of
booting, even if the kernel is being run at the address it is linked at,
since the linker does not necessarily fill in words in the image for
which there are dynamic relocations. (In fact the linker does fill in
such words for 64-bit executables, though not for 32-bit executables,
so in principle we could avoid calling relocate() entirely when we're
running a 64-bit kernel at the linked address.)
The dynamic relocations are processed by a new function relocate(addr),
where the addr parameter is the virtual address where the image will be
run. In fact we call it twice; once before calling prom_init, and again
when starting the main kernel. This means that reloc_offset() returns
0 in prom_init (since it has been relocated to the address it is running
at), which necessitated a few adjustments.
This also changes __va and __pa to use an equivalent definition that is
simpler. With the relocatable kernel, PAGE_OFFSET and MEMORY_START are
constants (for 64-bit) whereas PHYSICAL_START is a variable (and
KERNELBASE ideally should be too, but isn't yet).
With this, relocatable kernels still copy themselves down to physical
address 0 and run there.
Signed-off-by: Paul Mackerras <paulus@samba.org>
2008-08-30 11:43:47 +10:00
li r3 ,0 / * t a r g e t a d d r * /
2022-09-19 19:01:33 +02:00
# ifdef C O N F I G _ P P C _ B O O K 3 E _ 6 4
2015-10-06 22:48:14 -05:00
tovirt( r3 ,r3 ) / * o n b o o k e , w e a l r e a d y r u n a t P A G E _ O F F S E T * /
2009-07-23 23:15:59 +00:00
# endif
powerpc: Make the 64-bit kernel as a position-independent executable
This implements CONFIG_RELOCATABLE for 64-bit by making the kernel as
a position-independent executable (PIE) when it is set. This involves
processing the dynamic relocations in the image in the early stages of
booting, even if the kernel is being run at the address it is linked at,
since the linker does not necessarily fill in words in the image for
which there are dynamic relocations. (In fact the linker does fill in
such words for 64-bit executables, though not for 32-bit executables,
so in principle we could avoid calling relocate() entirely when we're
running a 64-bit kernel at the linked address.)
The dynamic relocations are processed by a new function relocate(addr),
where the addr parameter is the virtual address where the image will be
run. In fact we call it twice; once before calling prom_init, and again
when starting the main kernel. This means that reloc_offset() returns
0 in prom_init (since it has been relocated to the address it is running
at), which necessitated a few adjustments.
This also changes __va and __pa to use an equivalent definition that is
simpler. With the relocatable kernel, PAGE_OFFSET and MEMORY_START are
constants (for 64-bit) whereas PHYSICAL_START is a variable (and
KERNELBASE ideally should be too, but isn't yet).
With this, relocatable kernels still copy themselves down to physical
address 0 and run there.
Signed-off-by: Paul Mackerras <paulus@samba.org>
2008-08-30 11:43:47 +10:00
mr. r4 ,r26 / * I n s o m e c a s e s t h e l o a d e r m a y * /
2022-09-19 19:01:33 +02:00
# if d e f i n e d ( C O N F I G _ P P C _ B O O K 3 E _ 6 4 )
2015-10-06 22:48:14 -05:00
tovirt( r4 ,r4 )
# endif
2008-08-30 11:41:12 +10:00
beq 9 f / * h a v e a l r e a d y p u t u s a t z e r o * /
2005-09-26 16:04:21 +10:00
li r6 ,0 x10 0 / * S t a r t o f f s e t , t h e f i r s t 0 x10 0 * /
/* bytes were copied earlier. */
2012-11-11 19:01:05 +00:00
# ifdef C O N F I G _ R E L O C A T A B L E
2008-10-21 17:38:10 +00:00
/ *
* Check i f t h e k e r n e l h a s t o b e r u n n i n g a s r e l o c a t a b l e k e r n e l b a s e d o n t h e
2008-10-23 18:41:09 +00:00
* variable _ _ r u n _ a t _ l o a d , i f i t i s s e t t h e k e r n e l i s t r e a t e d a s r e l o c a t a b l e
2008-10-21 17:38:10 +00:00
* kernel, o t h e r w i s e i t w i l l b e m o v e d t o P H Y S I C A L _ S T A R T
* /
2022-09-19 19:01:33 +02:00
# if d e f i n e d ( C O N F I G _ P P C _ B O O K 3 E _ 6 4 )
2015-10-06 22:48:15 -05:00
tovirt( r26 ,r26 ) / * o n b o o k e , w e a l r e a d y r u n a t P A G E _ O F F S E T * /
# endif
2016-09-28 11:31:48 +10:00
lwz r7 ,( F I X E D _ S Y M B O L _ A B S _ A D D R ( _ _ r u n _ a t _ l o a d ) ) ( r26 )
2008-10-23 18:41:09 +00:00
cmplwi c r0 ,r7 ,1
2008-10-21 17:38:10 +00:00
bne 3 f
2022-09-19 19:01:33 +02:00
# ifdef C O N F I G _ P P C _ B O O K 3 E _ 6 4
2015-10-06 22:48:15 -05:00
LOAD_ R E G _ A D D R ( r5 , _ _ e n d _ i n t e r r u p t s )
LOAD_ R E G _ A D D R ( r11 , _ s t e x t )
sub r5 ,r5 ,r11
# else
2012-11-02 17:21:43 +11:00
/* just copy interrupts */
2019-08-20 14:34:14 +00:00
LOAD_ R E G _ I M M E D I A T E _ S Y M ( r5 , r11 , F I X E D _ S Y M B O L _ A B S _ A D D R ( _ _ e n d _ i n t e r r u p t s ) )
2015-10-06 22:48:15 -05:00
# endif
2008-10-21 17:38:10 +00:00
b 5 f
3 :
# endif
2016-09-28 11:31:48 +10:00
/* # bytes of memory to copy */
2021-12-21 16:59:02 +11:00
lis r5 ,( A B S _ A D D R ( c o p y _ t o _ h e r e , t e x t ) ) @ha
addi r5 ,r5 ,( A B S _ A D D R ( c o p y _ t o _ h e r e , t e x t ) ) @l
2008-10-21 17:38:10 +00:00
2014-02-04 16:04:35 +11:00
bl c o p y _ a n d _ f l u s h / * c o p y t h e f i r s t n b y t e s * /
2005-09-26 16:04:21 +10:00
/* this includes the code being */
/* executed here. */
2016-09-28 11:31:48 +10:00
/* Jump to the copy of this code that we just made */
2021-12-21 16:59:02 +11:00
addis r8 ,r3 ,( A B S _ A D D R ( 4 f , t e x t ) ) @ha
addi r12 ,r8 ,( A B S _ A D D R ( 4 f , t e x t ) ) @l
2014-02-04 16:07:47 +11:00
mtctr r12
2005-09-26 16:04:21 +10:00
bctr
2013-12-23 12:19:51 +11:00
.balign 8
2017-03-09 16:42:12 +11:00
p_end : .8byte _ e n d - c o p y _ t o _ h e r e
2008-10-21 17:38:10 +00:00
2016-09-28 11:31:47 +10:00
4 :
/ *
* Now c o p y t h e r e s t o f t h e k e r n e l u p t o _ e n d , a d d
* _ end - c o p y _ t o _ h e r e t o t h e c o p y l i m i t a n d r u n a g a i n .
* /
2021-12-21 16:59:02 +11:00
addis r8 ,r26 ,( A B S _ A D D R ( p _ e n d , t e x t ) ) @ha
ld r8 ,( A B S _ A D D R ( p _ e n d , t e x t ) ) @l(r8)
2016-09-28 11:31:47 +10:00
add r5 ,r5 ,r8
2014-02-04 16:04:35 +11:00
5 : bl c o p y _ a n d _ f l u s h / * c o p y t h e r e s t * /
2008-08-30 11:41:12 +10:00
2014-02-04 16:04:35 +11:00
9 : b s t a r t _ h e r e _ m u l t i p l a t f o r m
2008-08-30 11:41:12 +10:00
2005-09-26 16:04:21 +10:00
/ *
* Copy r o u t i n e u s e d t o c o p y t h e k e r n e l t o s t a r t a t p h y s i c a l a d d r e s s 0
* and f l u s h a n d i n v a l i d a t e t h e c a c h e s a s n e e d e d .
* r3 = d e s t a d d r , r4 = s o u r c e a d d r , r5 = c o p y l i m i t , r6 = s t a r t o f f s e t
* on e x i t , r3 , r4 , r5 a r e u n c h a n g e d , r6 i s u p d a t e d t o b e > = r5 .
*
* Note : this r o u t i n e * o n l y * c l o b b e r s r0 , r6 a n d l r
* /
_ GLOBAL( c o p y _ a n d _ f l u s h )
addi r5 ,r5 ,- 8
addi r6 ,r6 ,- 8
2006-09-06 14:34:41 -05:00
4 : li r0 ,8 / * U s e t h e s m a l l e s t c o m m o n * /
2005-09-26 16:04:21 +10:00
/* denominator cache line */
/* size. This results in */
/* extra cache line flushes */
/* but operation is correct. */
/* Can't get cache line size */
/* from NACA as it is being */
/* moved too. */
mtctr r0 / * p u t # w o r d s / l i n e i n c t r * /
3 : addi r6 ,r6 ,8 / * c o p y a c a c h e l i n e * /
ldx r0 ,r6 ,r4
stdx r0 ,r6 ,r3
bdnz 3 b
dcbst r6 ,r3 / * w r i t e i t t o m e m o r y * /
sync
icbi r6 ,r3 / * f l u s h t h e i c a c h e l i n e * /
cmpld 0 ,r6 ,r5
blt 4 b
sync
addi r5 ,r5 ,8
addi r6 ,r6 ,8
2013-04-24 00:30:09 +00:00
isync
2005-09-26 16:04:21 +10:00
blr
2021-07-14 18:17:58 +05:30
_ ASM_ N O K P R O B E _ S Y M B O L ( c o p y _ a n d _ f l u s h ) ; /* Called in real mode */
2005-09-26 16:04:21 +10:00
.align 8
copy_to_here :
# ifdef C O N F I G _ S M P
# ifdef C O N F I G _ P P C _ P M A C
/ *
* On P o w e r M a c , s e c o n d a r y p r o c e s s o r s s t a r t s f r o m t h e r e s e t v e c t o r , w h i c h
* is t e m p o r a r i l y t u r n e d i n t o a c a l l t o o n e o f t h e f u n c t i o n s b e l o w .
* /
.section " .text " ;
.align 2 ;
2005-10-22 16:02:39 +10:00
.globl __secondary_start_pmac_0
__secondary_start_pmac_0 :
/* NB the entries for cpus 0, 1, 2 must each occupy 8 bytes. */
li r24 ,0
b 1 f
li r24 ,1
b 1 f
li r24 ,2
b 1 f
li r24 ,3
1 :
2005-09-26 16:04:21 +10:00
_ GLOBAL( p m a c _ s e c o n d a r y _ s t a r t )
/* turn on 64-bit mode */
2014-02-04 16:04:35 +11:00
bl e n a b l e _ 6 4 b _ m o d e
2005-09-26 16:04:21 +10:00
2009-01-11 19:03:45 +00:00
li r0 ,0
mfspr r3 ,S P R N _ H I D 4
rldimi r3 ,r0 ,4 0 ,2 3 / * c l e a r b i t 2 3 ( r m _ c i ) * /
sync
mtspr S P R N _ H I D 4 ,r3
isync
sync
slbia
2008-08-30 11:41:12 +10:00
/* get TOC pointer (real address) */
2014-02-04 16:04:35 +11:00
bl r e l a t i v e _ t o c
2012-11-26 17:41:08 +00:00
tovirt( r2 ,r2 )
2008-08-30 11:41:12 +10:00
2005-09-26 16:04:21 +10:00
/* Copy some CPU settings from CPU 0 */
2014-02-04 16:04:35 +11:00
bl _ _ r e s t o r e _ c p u _ p p c97 0
2005-09-26 16:04:21 +10:00
/* pSeries do that early though I don't think we really need it */
mfmsr r3
ori r3 ,r3 ,M S R _ R I
mtmsrd r3 / * R I o n * /
/* Set up a paca value for this processor. */
2018-02-14 01:08:12 +10:00
LOAD_ R E G _ A D D R ( r4 ,p a c a _ p t r s ) / * L o a d p a c a p o i n t e r * /
ld r4 ,0 ( r4 ) / * G e t b a s e v a d d r o f p a c a _ p t r s a r r a y * /
sldi r5 ,r24 ,3 / * g e t p a c a _ p t r s [ ] i n d e x f r o m c p u i d * /
ldx r13 ,r5 ,r4 / * r13 = p a c a _ p t r s [ c p u i d ] * /
2011-01-20 17:50:21 +11:00
SET_ P A C A ( r13 ) / * S a v e v a d d r o f p a c a i n a n S P R G * /
2005-09-26 16:04:21 +10:00
2011-02-21 16:49:58 +11:00
/ * Mark i n t e r r u p t s s o f t a n d h a r d d i s a b l e d ( t h e y m i g h t b e e n a b l e d
* in t h e P A C A w h e n d o i n g h o t p l u g )
* /
2017-12-20 09:25:42 +05:30
li r0 ,I R Q S _ D I S A B L E D
2017-12-20 09:25:50 +05:30
stb r0 ,P A C A I R Q S O F T M A S K ( r13 )
powerpc: Rework lazy-interrupt handling
The current implementation of lazy interrupts handling has some
issues that this tries to address.
We don't do the various workarounds we need to do when re-enabling
interrupts in some cases such as when returning from an interrupt
and thus we may still lose or get delayed decrementer or doorbell
interrupts.
The current scheme also makes it much harder to handle the external
"edge" interrupts provided by some BookE processors when using the
EPR facility (External Proxy) and the Freescale Hypervisor.
Additionally, we tend to keep interrupts hard disabled in a number
of cases, such as decrementer interrupts, external interrupts, or
when a masked decrementer interrupt is pending. This is sub-optimal.
This is an attempt at fixing it all in one go by reworking the way
we do the lazy interrupt disabling from the ground up.
The base idea is to replace the "hard_enabled" field with a
"irq_happened" field in which we store a bit mask of what interrupt
occurred while soft-disabled.
When re-enabling, either via arch_local_irq_restore() or when returning
from an interrupt, we can now decide what to do by testing bits in that
field.
We then implement replaying of the missed interrupts either by
re-using the existing exception frame (in exception exit case) or via
the creation of a new one from an assembly trampoline (in the
arch_local_irq_enable case).
This removes the need to play with the decrementer to try to create
fake interrupts, among others.
In addition, this adds a few refinements:
- We no longer hard disable decrementer interrupts that occur
while soft-disabled. We now simply bump the decrementer back to max
(on BookS) or leave it stopped (on BookE) and continue with hard interrupts
enabled, which means that we'll potentially get better sample quality from
performance monitor interrupts.
- Timer, decrementer and doorbell interrupts now hard-enable
shortly after removing the source of the interrupt, which means
they no longer run entirely hard disabled. Again, this will improve
perf sample quality.
- On Book3E 64-bit, we now make the performance monitor interrupt
act as an NMI like Book3S (the necessary C code for that to work
appear to already be present in the FSL perf code, notably calling
nmi_enter instead of irq_enter). (This also fixes a bug where BookE
perfmon interrupts could clobber r14 ... oops)
- We could make "masked" decrementer interrupts act as NMIs when doing
timer-based perf sampling to improve the sample quality.
Signed-off-by-yet: Benjamin Herrenschmidt <benh@kernel.crashing.org>
---
v2:
- Add hard-enable to decrementer, timer and doorbells
- Fix CR clobber in masked irq handling on BookE
- Make embedded perf interrupt act as an NMI
- Add a PACA_HAPPENED_EE_EDGE for use by FSL if they want
to retrigger an interrupt without preventing hard-enable
v3:
- Fix or vs. ori bug on Book3E
- Fix enabling of interrupts for some exceptions on Book3E
v4:
- Fix resend of doorbells on return from interrupt on Book3E
v5:
- Rebased on top of my latest series, which involves some significant
rework of some aspects of the patch.
v6:
- 32-bit compile fix
- more compile fixes with various .config combos
- factor out the asm code to soft-disable interrupts
- remove the C wrapper around preempt_schedule_irq
v7:
- Fix a bug with hard irq state tracking on native power7
2012-03-06 18:27:59 +11:00
li r0 ,P A C A _ I R Q _ H A R D _ D I S
stb r0 ,P A C A I R Q H A P P E N E D ( r13 )
2011-02-21 16:49:58 +11:00
2005-09-26 16:04:21 +10:00
/* Create a temp kernel stack for use before relocation is on. */
ld r1 ,P A C A E M E R G S P ( r13 )
2022-11-27 22:49:40 +10:00
subi r1 ,r1 ,S T A C K _ F R A M E _ M I N _ S I Z E
2005-09-26 16:04:21 +10:00
2006-11-27 14:59:50 +11:00
b _ _ s e c o n d a r y _ s t a r t
2005-09-26 16:04:21 +10:00
# endif / * C O N F I G _ P P C _ P M A C * /
/ *
* This f u n c t i o n i s c a l l e d a f t e r t h e m a s t e r C P U h a s r e l e a s e d t h e
* secondary p r o c e s s o r s . T h e e x e c u t i o n e n v i r o n m e n t i s r e l o c a t i o n o f f .
* The p a c a f o r t h i s p r o c e s s o r h a s t h e f o l l o w i n g f i e l d s i n i t i a l i z e d a t
* this p o i n t :
* 1 . Processor n u m b e r
* 2 . Segment t a b l e p o i n t e r ( v i r t u a l a d d r e s s )
* On e n t r y t h e f o l l o w i n g a r e s e t :
2012-02-28 13:44:58 +11:00
* r1 = s t a c k p o i n t e r ( r e a l a d d r o f t e m p s t a c k )
2009-07-14 20:52:54 +00:00
* r2 4 = c p u # ( i n L i n u x t e r m s )
* r1 3 = p a c a v i r t u a l a d d r e s s
* SPRG_ P A C A = p a c a v i r t u a l a d d r e s s
2005-09-26 16:04:21 +10:00
* /
2009-07-23 23:15:59 +00:00
.section " .text " ;
.align 2 ;
2007-08-22 13:44:58 +10:00
.globl __secondary_start
2006-11-27 14:59:50 +11:00
__secondary_start :
2005-11-10 13:37:51 +11:00
/* Set thread priority to MEDIUM */
HMT_ M E D I U M
2005-09-26 16:04:21 +10:00
2019-02-22 23:58:21 +11:00
/ *
* Do e a r l y s e t u p f o r t h i s C P U , i n p a r t i c u l a r i n i t i a l i s i n g t h e M M U s o w e
* can t u r n i t o n b e l o w . T h i s i s a c a l l t o C , w h i c h i s O K , w e ' r e s t i l l
* running o n t h e e m e r g e n c y s t a c k .
* /
2014-02-04 16:04:35 +11:00
bl e a r l y _ s e t u p _ s e c o n d a r y
2010-08-12 20:58:28 +00:00
2010-08-25 21:04:25 +00:00
/ *
2019-02-22 23:58:21 +11:00
* The p r i m a r y h a s i n i t i a l i z e d o u r k e r n e l s t a c k f o r u s i n t h e p a c a , g r a b
* it a n d p u t i t i n r1 . W e m u s t * n o t * u s e i t u n t i l w e t u r n o n t h e M M U
* below, b e c a u s e i t m a y n o t b e i n s i d e t h e R M O .
2010-08-25 21:04:25 +00:00
* /
2019-02-22 23:58:21 +11:00
ld r1 , P A C A K S A V E ( r13 )
2010-08-25 21:04:25 +00:00
2005-11-10 13:37:51 +11:00
/* Clear backchain so we get nice backtraces */
2005-09-26 16:04:21 +10:00
li r7 ,0
mtlr r7
powerpc: Rework lazy-interrupt handling
The current implementation of lazy interrupts handling has some
issues that this tries to address.
We don't do the various workarounds we need to do when re-enabling
interrupts in some cases such as when returning from an interrupt
and thus we may still lose or get delayed decrementer or doorbell
interrupts.
The current scheme also makes it much harder to handle the external
"edge" interrupts provided by some BookE processors when using the
EPR facility (External Proxy) and the Freescale Hypervisor.
Additionally, we tend to keep interrupts hard disabled in a number
of cases, such as decrementer interrupts, external interrupts, or
when a masked decrementer interrupt is pending. This is sub-optimal.
This is an attempt at fixing it all in one go by reworking the way
we do the lazy interrupt disabling from the ground up.
The base idea is to replace the "hard_enabled" field with a
"irq_happened" field in which we store a bit mask of what interrupt
occurred while soft-disabled.
When re-enabling, either via arch_local_irq_restore() or when returning
from an interrupt, we can now decide what to do by testing bits in that
field.
We then implement replaying of the missed interrupts either by
re-using the existing exception frame (in exception exit case) or via
the creation of a new one from an assembly trampoline (in the
arch_local_irq_enable case).
This removes the need to play with the decrementer to try to create
fake interrupts, among others.
In addition, this adds a few refinements:
- We no longer hard disable decrementer interrupts that occur
while soft-disabled. We now simply bump the decrementer back to max
(on BookS) or leave it stopped (on BookE) and continue with hard interrupts
enabled, which means that we'll potentially get better sample quality from
performance monitor interrupts.
- Timer, decrementer and doorbell interrupts now hard-enable
shortly after removing the source of the interrupt, which means
they no longer run entirely hard disabled. Again, this will improve
perf sample quality.
- On Book3E 64-bit, we now make the performance monitor interrupt
act as an NMI like Book3S (the necessary C code for that to work
appear to already be present in the FSL perf code, notably calling
nmi_enter instead of irq_enter). (This also fixes a bug where BookE
perfmon interrupts could clobber r14 ... oops)
- We could make "masked" decrementer interrupts act as NMIs when doing
timer-based perf sampling to improve the sample quality.
Signed-off-by-yet: Benjamin Herrenschmidt <benh@kernel.crashing.org>
---
v2:
- Add hard-enable to decrementer, timer and doorbells
- Fix CR clobber in masked irq handling on BookE
- Make embedded perf interrupt act as an NMI
- Add a PACA_HAPPENED_EE_EDGE for use by FSL if they want
to retrigger an interrupt without preventing hard-enable
v3:
- Fix or vs. ori bug on Book3E
- Fix enabling of interrupts for some exceptions on Book3E
v4:
- Fix resend of doorbells on return from interrupt on Book3E
v5:
- Rebased on top of my latest series, which involves some significant
rework of some aspects of the patch.
v6:
- 32-bit compile fix
- more compile fixes with various .config combos
- factor out the asm code to soft-disable interrupts
- remove the C wrapper around preempt_schedule_irq
v7:
- Fix a bug with hard irq state tracking on native power7
2012-03-06 18:27:59 +11:00
/ * Mark i n t e r r u p t s s o f t a n d h a r d d i s a b l e d ( t h e y m i g h t b e e n a b l e d
* in t h e P A C A w h e n d o i n g h o t p l u g )
* /
2017-12-20 09:25:42 +05:30
li r7 ,I R Q S _ D I S A B L E D
2017-12-20 09:25:50 +05:30
stb r7 ,P A C A I R Q S O F T M A S K ( r13 )
powerpc: Rework lazy-interrupt handling
The current implementation of lazy interrupts handling has some
issues that this tries to address.
We don't do the various workarounds we need to do when re-enabling
interrupts in some cases such as when returning from an interrupt
and thus we may still lose or get delayed decrementer or doorbell
interrupts.
The current scheme also makes it much harder to handle the external
"edge" interrupts provided by some BookE processors when using the
EPR facility (External Proxy) and the Freescale Hypervisor.
Additionally, we tend to keep interrupts hard disabled in a number
of cases, such as decrementer interrupts, external interrupts, or
when a masked decrementer interrupt is pending. This is sub-optimal.
This is an attempt at fixing it all in one go by reworking the way
we do the lazy interrupt disabling from the ground up.
The base idea is to replace the "hard_enabled" field with a
"irq_happened" field in which we store a bit mask of what interrupt
occurred while soft-disabled.
When re-enabling, either via arch_local_irq_restore() or when returning
from an interrupt, we can now decide what to do by testing bits in that
field.
We then implement replaying of the missed interrupts either by
re-using the existing exception frame (in exception exit case) or via
the creation of a new one from an assembly trampoline (in the
arch_local_irq_enable case).
This removes the need to play with the decrementer to try to create
fake interrupts, among others.
In addition, this adds a few refinements:
- We no longer hard disable decrementer interrupts that occur
while soft-disabled. We now simply bump the decrementer back to max
(on BookS) or leave it stopped (on BookE) and continue with hard interrupts
enabled, which means that we'll potentially get better sample quality from
performance monitor interrupts.
- Timer, decrementer and doorbell interrupts now hard-enable
shortly after removing the source of the interrupt, which means
they no longer run entirely hard disabled. Again, this will improve
perf sample quality.
- On Book3E 64-bit, we now make the performance monitor interrupt
act as an NMI like Book3S (the necessary C code for that to work
appear to already be present in the FSL perf code, notably calling
nmi_enter instead of irq_enter). (This also fixes a bug where BookE
perfmon interrupts could clobber r14 ... oops)
- We could make "masked" decrementer interrupts act as NMIs when doing
timer-based perf sampling to improve the sample quality.
Signed-off-by-yet: Benjamin Herrenschmidt <benh@kernel.crashing.org>
---
v2:
- Add hard-enable to decrementer, timer and doorbells
- Fix CR clobber in masked irq handling on BookE
- Make embedded perf interrupt act as an NMI
- Add a PACA_HAPPENED_EE_EDGE for use by FSL if they want
to retrigger an interrupt without preventing hard-enable
v3:
- Fix or vs. ori bug on Book3E
- Fix enabling of interrupts for some exceptions on Book3E
v4:
- Fix resend of doorbells on return from interrupt on Book3E
v5:
- Rebased on top of my latest series, which involves some significant
rework of some aspects of the patch.
v6:
- 32-bit compile fix
- more compile fixes with various .config combos
- factor out the asm code to soft-disable interrupts
- remove the C wrapper around preempt_schedule_irq
v7:
- Fix a bug with hard irq state tracking on native power7
2012-03-06 18:27:59 +11:00
li r0 ,P A C A _ I R Q _ H A R D _ D I S
stb r0 ,P A C A I R Q H A P P E N E D ( r13 )
2012-02-28 13:44:58 +11:00
2005-09-26 16:04:21 +10:00
/* enable MMU and jump to start_secondary */
2014-02-04 16:04:52 +11:00
LOAD_ R E G _ A D D R ( r3 , s t a r t _ s e c o n d a r y _ p r o l o g )
2006-01-13 14:56:25 +11:00
LOAD_ R E G _ I M M E D I A T E ( r4 , M S R _ K E R N E L )
[POWERPC] Lazy interrupt disabling for 64-bit machines
This implements a lazy strategy for disabling interrupts. This means
that local_irq_disable() et al. just clear the 'interrupts are
enabled' flag in the paca. If an interrupt comes along, the interrupt
entry code notices that interrupts are supposed to be disabled, and
clears the EE bit in SRR1, clears the 'interrupts are hard-enabled'
flag in the paca, and returns. This means that interrupts only
actually get disabled in the processor when an interrupt comes along.
When interrupts are enabled by local_irq_enable() et al., the code
sets the interrupts-enabled flag in the paca, and then checks whether
interrupts got hard-disabled. If so, it also sets the EE bit in the
MSR to hard-enable the interrupts.
This has the potential to improve performance, and also makes it
easier to make a kernel that can boot on iSeries and on other 64-bit
machines, since this lazy-disable strategy is very similar to the
soft-disable strategy that iSeries already uses.
This version renames paca->proc_enabled to paca->soft_enabled, and
changes a couple of soft-disables in the kexec code to hard-disables,
which should fix the crash that Michael Ellerman saw. This doesn't
yet use a reserved CR field for the soft_enabled and hard_enabled
flags. This applies on top of Stephen Rothwell's patches to make it
possible to build a combined iSeries/other kernel.
Signed-off-by: Paul Mackerras <paulus@samba.org>
2006-10-04 16:47:49 +10:00
2005-10-10 14:01:07 +10:00
mtspr S P R N _ S R R 0 ,r3
mtspr S P R N _ S R R 1 ,r4
2020-11-08 16:57:35 +00:00
RFI_ T O _ K E R N E L
2005-09-26 16:04:21 +10:00
b . / * p r e v e n t s p e c u l a t i v e e x e c u t i o n * /
/ *
* Running w i t h r e l o c a t i o n o n a t t h i s p o i n t . A l l w e w a n t t o d o i s
2008-08-30 11:41:12 +10:00
* zero t h e s t a c k b a c k - c h a i n p o i n t e r a n d g e t t h e T O C v i r t u a l a d d r e s s
* before g o i n g i n t o C c o d e .
2005-09-26 16:04:21 +10:00
* /
2014-02-04 16:04:52 +11:00
start_secondary_prolog :
2022-09-26 13:40:56 +10:00
LOAD_ P A C A _ T O C ( )
2005-09-26 16:04:21 +10:00
li r3 ,0
std r3 ,0 ( r1 ) / * Z e r o t h e s t a c k f r a m e p o i n t e r * /
2014-02-04 16:04:35 +11:00
bl s t a r t _ s e c o n d a r y
2005-11-10 13:37:51 +11:00
b .
2010-03-01 02:58:09 +00:00
/ *
* Reset s t a c k p o i n t e r a n d c a l l s t a r t _ s e c o n d a r y
* to c o n t i n u e w i t h o n l i n e o p e r a t i o n w h e n w o k e n u p
* from c e d e i n c p u o f f l i n e .
* /
_ GLOBAL( s t a r t _ s e c o n d a r y _ r e s u m e )
ld r1 ,P A C A K S A V E ( r13 ) / * R e l o a d k e r n e l s t a c k p o i n t e r * /
li r3 ,0
std r3 ,0 ( r1 ) / * Z e r o t h e s t a c k f r a m e p o i n t e r * /
2014-02-04 16:04:35 +11:00
bl s t a r t _ s e c o n d a r y
2010-03-01 02:58:09 +00:00
b .
2005-09-26 16:04:21 +10:00
# endif
/ *
* This s u b r o u t i n e c l o b b e r s r11 a n d r12
* /
2022-11-14 23:27:41 +05:30
SYM_ F U N C _ S T A R T _ L O C A L ( e n a b l e _ 6 4 b _ m o d e )
2005-09-26 16:04:21 +10:00
mfmsr r11 / * g r a b t h e c u r r e n t M S R * /
2022-09-19 19:01:33 +02:00
# ifdef C O N F I G _ P P C _ B O O K 3 E _ 6 4
2009-07-23 23:15:59 +00:00
oris r11 ,r11 ,0 x80 0 0 / * C M b i t s e t , w e ' l l s e t I C M l a t e r * /
mtmsr r11
2022-09-19 19:01:33 +02:00
# else / * C O N F I G _ P P C _ B O O K 3 E _ 6 4 * /
2020-11-06 14:53:40 +10:00
LOAD_ R E G _ I M M E D I A T E ( r12 , M S R _ 6 4 B I T )
2005-09-26 16:04:21 +10:00
or r11 ,r11 ,r12
mtmsrd r11
isync
2009-07-23 23:15:59 +00:00
# endif
2005-09-26 16:04:21 +10:00
blr
2022-11-14 23:27:41 +05:30
SYM_ F U N C _ E N D ( e n a b l e _ 6 4 b _ m o d e )
2005-09-26 16:04:21 +10:00
2008-08-30 11:41:12 +10:00
/ *
* This p u t s t h e T O C p o i n t e r i n t o r2 , o f f s e t b y 0 x80 0 0 ( a s e x p e c t e d
* by t h e t o o l c h a i n ) . I t c o m p u t e s t h e c o r r e c t v a l u e f o r w h e r e v e r w e
* are r u n n i n g a t t h e m o m e n t , u s i n g p o s i t i o n - i n d e p e n d e n t c o d e .
2012-11-26 17:41:08 +00:00
*
* Note : The c o m p i l e r c o n s t r u c t s p o i n t e r s u s i n g o f f s e t s f r o m t h e
* TOC i n - m c m o d e l =medium m o d e . A f t e r w e r e l o c a t e t o 0 b u t b e f o r e
* the M M U i s o n w e n e e d o u r T O C t o b e a v i r t u a l a d d r e s s o t h e r w i s e
* these p o i n t e r s w i l l b e r e a l a d d r e s s e s w h i c h m a y g e t s t o r e d a n d
* accessed l a t e r w i t h t h e M M U o n . W e u s e t o v i r t ( ) a t t h e c a l l
* sites t o h a n d l e t h i s .
2008-08-30 11:41:12 +10:00
* /
_ GLOBAL( r e l a t i v e _ t o c )
mflr r0
bcl 2 0 ,3 1 ,$ + 4
2011-09-19 17:44:51 +00:00
0 : mflr r11
ld r2 ,( p _ t o c - 0 b ) ( r11 )
add r2 ,r2 ,r11
2008-08-30 11:41:12 +10:00
mtlr r0
blr
2013-08-07 02:01:18 +10:00
.balign 8
powerpc/toc: Future proof kernel toc
This patch future-proofs the kernel against linker changes that might
put the toc pointer at some location other than .got+0x8000, by
replacing __toc_start+0x8000 with .TOC. throughout. If the kernel's
idea of the toc pointer doesn't agree with the linker, bad things
happen.
prom_init.c code relocating its toc is also changed so that a symbolic
__prom_init_toc_start toc-pointer relative address is calculated
rather than assuming that it is always at toc-pointer - 0x8000. The
length calculations loading values from the toc are also avoided.
It's a little incestuous to do that with unreloc_toc picking up
adjusted values (which is fine in practice, they both adjust by the
same amount if all goes well).
I've also changed the way .got is aligned in vmlinux.lds and
zImage.lds, mostly so that dumping out section info by objdump or
readelf plainly shows the alignment is 256. This linker script
feature was added 2005-09-27, available in FSF binutils releases from
2.17 onwards. Should be safe to use in the kernel, I think.
Finally, put *(.got) before the prom_init.o entry which only needs
*(.toc), so that the GOT header goes in the correct place. I don't
believe this makes any difference for the kernel as it would for
dynamic objects being loaded by ld.so. That change is just to stop
lusers who blindly copy kernel scripts being led astray. Of course,
this change needs the prom_init.c changes.
Some notes on .toc and .got.
.toc is a compiler generated section of addresses. .got is a linker
generated section of addresses, generally built when the linker sees
R_*_*GOT* relocations. In the case of powerpc64 ld.bfd, there are
multiple generated .got sections, one per input object file. So you
can somewhat reasonably write in a linker script an input section
statement like *prom_init.o(.got .toc) to mean "the .got and .toc
section for files matching *prom_init.o". On other architectures that
doesn't make sense, because the linker generally has just one .got
section. Even on powerpc64, note well that the GOT entries for
prom_init.o may be merged with GOT entries from other objects. That
means that if prom_init.o references, say, _end via some GOT
relocation, and some other object also references _end via a GOT
relocation, the GOT entry for _end may be in the range
__prom_init_toc_start to __prom_init_toc_end and if the kernel does
something special to GOT/TOC entries in that range then the value of
_end as seen by objects other than prom_init.o will be affected. On
the other hand the GOT entry for _end may not be in the range
__prom_init_toc_start to __prom_init_toc_end. Which way it turns out
is deterministic but a detail of linker operation that should not be
relied on.
A feature of ld.bfd is that input .toc (and .got) sections matching
one linker input section statement may be sorted, to put entries used
by small-model code first, near the toc base. This is why scripts for
powerpc64 normally use *(.got .toc) rather than *(.got) *(.toc), since
the first form allows more freedom to sort.
Another feature of ld.bfd is that indirect addressing sequences using
the GOT/TOC may be edited by the linker to relative addressing. In
many cases relative addressing would be emitted by gcc for
-mcmodel=medium if you appropriately decorate variable declarations
with non-default visibility.
The original patch is here:
https://lore.kernel.org/linuxppc-dev/20210310034813.GM6042@bubble.grove.modra.org/
Signed-off-by: Alan Modra <amodra@au1.ibm.com>
[aik: removed non-relocatable which is gone in 24d33ac5b8ffb]
[aik: added <=2.24 check]
[aik: because of llvm-as, kernel_toc_addr() uses "mr" instead of global register variable]
Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20211221055904.555763-2-aik@ozlabs.ru
2021-12-21 16:58:59 +11:00
p_toc : .8byte . T O C . - 0 b
2008-08-30 11:41:12 +10:00
2005-09-26 16:04:21 +10:00
/ *
* This i s w h e r e t h e m a i n k e r n e l c o d e s t a r t s .
* /
2019-05-10 06:31:28 +00:00
_ _ REF
2014-02-04 16:06:11 +11:00
start_here_multiplatform :
2012-11-26 17:41:08 +00:00
/* set up the TOC */
2014-02-04 16:04:35 +11:00
bl r e l a t i v e _ t o c
2012-11-26 17:41:08 +00:00
tovirt( r2 ,r2 )
2005-09-26 16:04:21 +10:00
/ * Clear o u t t h e B S S . I t m a y h a v e b e e n d o n e i n p r o m _ i n i t ,
* already b u t t h a t ' s i r r e l e v a n t s i n c e p r o m _ i n i t w i l l s o o n
* be d e t a c h e d f r o m t h e k e r n e l c o m p l e t e l y . B e s i d e s , w e n e e d
* to c l e a r i t n o w f o r k e x e c - s t y l e e n t r y .
* /
2008-08-30 11:41:12 +10:00
LOAD_ R E G _ A D D R ( r11 ,_ _ b s s _ s t o p )
LOAD_ R E G _ A D D R ( r8 ,_ _ b s s _ s t a r t )
2005-09-26 16:04:21 +10:00
sub r11 ,r11 ,r8 / * b s s s i z e * /
addi r11 ,r11 ,7 / * r o u n d u p t o a n e v e n d o u b l e w o r d * /
2008-08-30 11:41:12 +10:00
srdi. r11 ,r11 ,3 / * s h i f t r i g h t b y 3 * /
2005-09-26 16:04:21 +10:00
beq 4 f
addi r8 ,r8 ,- 8
li r0 ,0
mtctr r11 / * z e r o t h i s m a n y d o u b l e w o r d s * /
3 : stdu r0 ,8 ( r8 )
bdnz 3 b
4 :
2011-09-19 17:44:59 +00:00
# ifdef C O N F I G _ P P C _ E A R L Y _ D E B U G _ O P A L
/* Setup OPAL entry */
2012-10-21 14:30:52 +00:00
LOAD_ R E G _ A D D R ( r11 , o p a l )
2011-09-19 17:44:59 +00:00
std r28 ,0 ( r11 ) ;
std r29 ,8 ( r11 ) ;
# endif
2022-09-19 19:01:33 +02:00
# ifndef C O N F I G _ P P C _ B O O K 3 E _ 6 4
2005-09-26 16:04:21 +10:00
mfmsr r6
ori r6 ,r6 ,M S R _ R I
mtmsrd r6 / * R I o n * /
2009-07-23 23:15:59 +00:00
# endif
2005-09-26 16:04:21 +10:00
powerpc: Make the 64-bit kernel as a position-independent executable
This implements CONFIG_RELOCATABLE for 64-bit by making the kernel as
a position-independent executable (PIE) when it is set. This involves
processing the dynamic relocations in the image in the early stages of
booting, even if the kernel is being run at the address it is linked at,
since the linker does not necessarily fill in words in the image for
which there are dynamic relocations. (In fact the linker does fill in
such words for 64-bit executables, though not for 32-bit executables,
so in principle we could avoid calling relocate() entirely when we're
running a 64-bit kernel at the linked address.)
The dynamic relocations are processed by a new function relocate(addr),
where the addr parameter is the virtual address where the image will be
run. In fact we call it twice; once before calling prom_init, and again
when starting the main kernel. This means that reloc_offset() returns
0 in prom_init (since it has been relocated to the address it is running
at), which necessitated a few adjustments.
This also changes __va and __pa to use an equivalent definition that is
simpler. With the relocatable kernel, PAGE_OFFSET and MEMORY_START are
constants (for 64-bit) whereas PHYSICAL_START is a variable (and
KERNELBASE ideally should be too, but isn't yet).
With this, relocatable kernels still copy themselves down to physical
address 0 and run there.
Signed-off-by: Paul Mackerras <paulus@samba.org>
2008-08-30 11:43:47 +10:00
# ifdef C O N F I G _ R E L O C A T A B L E
/* Save the physical address we're running at in kernstart_addr */
LOAD_ R E G _ A D D R ( r4 , k e r n s t a r t _ a d d r )
clrldi r0 ,r25 ,2
std r0 ,0 ( r4 )
# endif
2008-08-30 11:41:12 +10:00
/* set up a stack pointer */
powerpc/64: Don't initialise init_task->thread.regs
Aneesh increased the size of struct pt_regs by 16 bytes and started
seeing this WARN_ON:
smp: Bringing up secondary CPUs ...
------------[ cut here ]------------
WARNING: CPU: 0 PID: 0 at arch/powerpc/kernel/process.c:455 giveup_all+0xb4/0x110
Modules linked in:
CPU: 0 PID: 0 Comm: swapper/0 Not tainted 5.7.0-rc2-gcc-8.2.0-1.g8f6a41f-default+ #318
NIP: c00000000001a2b4 LR: c00000000001a29c CTR: c0000000031d0000
REGS: c0000000026d3980 TRAP: 0700 Not tainted (5.7.0-rc2-gcc-8.2.0-1.g8f6a41f-default+)
MSR: 800000000282b033 <SF,VEC,VSX,EE,FP,ME,IR,DR,RI,LE> CR: 48048224 XER: 00000000
CFAR: c000000000019cc8 IRQMASK: 1
GPR00: c00000000001a264 c0000000026d3c20 c0000000026d7200 800000000280b033
GPR04: 0000000000000001 0000000000000000 0000000000000077 30206d7372203164
GPR08: 0000000000002000 0000000002002000 800000000280b033 3230303030303030
GPR12: 0000000000008800 c0000000031d0000 0000000000800050 0000000002000066
GPR16: 000000000309a1a0 000000000309a4b0 000000000309a2d8 000000000309a890
GPR20: 00000000030d0098 c00000000264da40 00000000fd620000 c0000000ff798080
GPR24: c00000000264edf0 c0000001007469f0 00000000fd620000 c0000000020e5e90
GPR28: c00000000264edf0 c00000000264d200 000000001db60000 c00000000264d200
NIP [c00000000001a2b4] giveup_all+0xb4/0x110
LR [c00000000001a29c] giveup_all+0x9c/0x110
Call Trace:
[c0000000026d3c20] [c00000000001a264] giveup_all+0x64/0x110 (unreliable)
[c0000000026d3c90] [c00000000001ae34] __switch_to+0x104/0x480
[c0000000026d3cf0] [c000000000e0b8a0] __schedule+0x320/0x970
[c0000000026d3dd0] [c000000000e0c518] schedule_idle+0x38/0x70
[c0000000026d3df0] [c00000000019c7c8] do_idle+0x248/0x3f0
[c0000000026d3e70] [c00000000019cbb8] cpu_startup_entry+0x38/0x40
[c0000000026d3ea0] [c000000000011bb0] rest_init+0xe0/0xf8
[c0000000026d3ed0] [c000000002004820] start_kernel+0x990/0x9e0
[c0000000026d3f90] [c00000000000c49c] start_here_common+0x1c/0x400
Which was unexpected. The warning is checking the thread.regs->msr
value of the task we are switching from:
usermsr = tsk->thread.regs->msr;
...
WARN_ON((usermsr & MSR_VSX) && !((usermsr & MSR_FP) && (usermsr & MSR_VEC)));
ie. if MSR_VSX is set then both of MSR_FP and MSR_VEC are also set.
Dumping tsk->thread.regs->msr we see that it's: 0x1db60000
Which is not a normal looking MSR, in fact the only valid bit is
MSR_VSX, all the other bits are reserved in the current definition of
the MSR.
We can see from the oops that it was swapper/0 that we were switching
from when we hit the warning, ie. init_task. So its thread.regs points
to the base (high addresses) in init_stack.
Dumping the content of init_task->thread.regs, with the members of
pt_regs annotated (the 16 bytes larger version), we see:
0000000000000000 c000000002780080 gpr[0] gpr[1]
0000000000000000 c000000002666008 gpr[2] gpr[3]
c0000000026d3ed0 0000000000000078 gpr[4] gpr[5]
c000000000011b68 c000000002780080 gpr[6] gpr[7]
0000000000000000 0000000000000000 gpr[8] gpr[9]
c0000000026d3f90 0000800000002200 gpr[10] gpr[11]
c000000002004820 c0000000026d7200 gpr[12] gpr[13]
000000001db60000 c0000000010aabe8 gpr[14] gpr[15]
c0000000010aabe8 c0000000010aabe8 gpr[16] gpr[17]
c00000000294d598 0000000000000000 gpr[18] gpr[19]
0000000000000000 0000000000001ff8 gpr[20] gpr[21]
0000000000000000 c00000000206d608 gpr[22] gpr[23]
c00000000278e0cc 0000000000000000 gpr[24] gpr[25]
000000002fff0000 c000000000000000 gpr[26] gpr[27]
0000000002000000 0000000000000028 gpr[28] gpr[29]
000000001db60000 0000000004750000 gpr[30] gpr[31]
0000000002000000 000000001db60000 nip msr
0000000000000000 0000000000000000 orig_r3 ctr
c00000000000c49c 0000000000000000 link xer
0000000000000000 0000000000000000 ccr softe
0000000000000000 0000000000000000 trap dar
0000000000000000 0000000000000000 dsisr result
0000000000000000 0000000000000000 ppr kuap
0000000000000000 0000000000000000 pad[2] pad[3]
This looks suspiciously like stack frames, not a pt_regs. If we look
closely we can see return addresses from the stack trace above,
c000000002004820 (start_kernel) and c00000000000c49c (start_here_common).
init_task->thread.regs is setup at build time in processor.h:
#define INIT_THREAD { \
.ksp = INIT_SP, \
.regs = (struct pt_regs *)INIT_SP - 1, /* XXX bogus, I think */ \
The early boot code where we setup the initial stack is:
LOAD_REG_ADDR(r3,init_thread_union)
/* set up a stack pointer */
LOAD_REG_IMMEDIATE(r1,THREAD_SIZE)
add r1,r3,r1
li r0,0
stdu r0,-STACK_FRAME_OVERHEAD(r1)
Which creates a stack frame of size 112 bytes (STACK_FRAME_OVERHEAD).
Which is far too small to contain a pt_regs.
So the result is init_task->thread.regs is pointing at some stack
frames on the init stack, not at a pt_regs.
We have gotten away with this for so long because with pt_regs at its
current size the MSR happens to point into the first frame, at a
location that is not written to by the early asm. With the 16 byte
expansion the MSR falls into the second frame, which is used by the
compiler, and collides with a saved register that tends to be
non-zero.
As far as I can see this has been wrong since the original merge of
64-bit ppc support, back in 2002.
Conceptually swapper should have no regs, it never entered from
userspace, and in fact that's what we do on 32-bit. It's also
presumably what the "bogus" comment is referring to.
So I think the right fix is to just not-initialise regs at all. I'm
slightly worried this will break some code that isn't prepared for a
NULL regs, but we'll have to see.
Remove the comment in head_64.S which refers to us setting up the
regs (even though we never did), and is otherwise not really accurate
any more.
Reported-by: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20200428123130.73078-1-mpe@ellerman.id.au
2020-04-28 22:31:30 +10:00
LOAD_ R E G _ A D D R ( r3 ,i n i t _ t h r e a d _ u n i o n )
2017-02-24 13:52:10 +13:00
LOAD_ R E G _ I M M E D I A T E ( r1 ,T H R E A D _ S I Z E )
add r1 ,r3 ,r1
2005-09-26 16:04:21 +10:00
li r0 ,0
2022-11-27 22:49:40 +10:00
stdu r0 ,- S T A C K _ F R A M E _ M I N _ S I Z E ( r1 )
2005-09-26 16:04:21 +10:00
2014-07-10 12:29:19 +10:00
/ *
* Do v e r y e a r l y k e r n e l i n i t i a l i z a t i o n s , i n c l u d i n g i n i t i a l h a s h t a b l e
* and S L B s e t u p b e f o r e w e t u r n o n r e l o c a t i o n .
* /
2005-09-26 16:04:21 +10:00
2022-06-28 16:48:59 +02:00
# ifdef C O N F I G _ K A S A N
bl k a s a n _ e a r l y _ i n i t
# endif
2005-09-26 16:04:21 +10:00
/* Restore parameters passed from prom_init/kexec */
mr r3 ,r31
2019-03-27 14:35:54 +11:00
LOAD_ R E G _ A D D R ( r12 , D O T S Y M ( e a r l y _ s e t u p ) )
mtctr r12
bctrl / * a l s o s e t s r13 a n d S P R G _ P A C A * /
2005-09-26 16:04:21 +10:00
2014-02-04 16:04:52 +11:00
LOAD_ R E G _ A D D R ( r3 , s t a r t _ h e r e _ c o m m o n )
2008-08-30 11:41:12 +10:00
ld r4 ,P A C A K M S R ( r13 )
2005-10-10 14:01:07 +10:00
mtspr S P R N _ S R R 0 ,r3
mtspr S P R N _ S R R 1 ,r4
2020-11-08 16:57:35 +00:00
RFI_ T O _ K E R N E L
2005-09-26 16:04:21 +10:00
b . / * p r e v e n t s p e c u l a t i v e e x e c u t i o n * /
2016-07-05 15:04:09 +10:00
2005-09-26 16:04:21 +10:00
/* This is where all platforms converge execution */
2014-02-04 16:04:52 +11:00
start_here_common :
2005-09-26 16:04:21 +10:00
/* relocation is on at this point */
2008-08-30 11:41:12 +10:00
std r1 ,P A C A K S A V E ( r13 )
2005-09-26 16:04:21 +10:00
2008-08-30 11:41:12 +10:00
/* Load the TOC (virtual address) */
2022-09-26 13:40:56 +10:00
LOAD_ P A C A _ T O C ( )
2005-09-26 16:04:21 +10:00
powerpc: Rework lazy-interrupt handling
The current implementation of lazy interrupts handling has some
issues that this tries to address.
We don't do the various workarounds we need to do when re-enabling
interrupts in some cases such as when returning from an interrupt
and thus we may still lose or get delayed decrementer or doorbell
interrupts.
The current scheme also makes it much harder to handle the external
"edge" interrupts provided by some BookE processors when using the
EPR facility (External Proxy) and the Freescale Hypervisor.
Additionally, we tend to keep interrupts hard disabled in a number
of cases, such as decrementer interrupts, external interrupts, or
when a masked decrementer interrupt is pending. This is sub-optimal.
This is an attempt at fixing it all in one go by reworking the way
we do the lazy interrupt disabling from the ground up.
The base idea is to replace the "hard_enabled" field with a
"irq_happened" field in which we store a bit mask of what interrupt
occurred while soft-disabled.
When re-enabling, either via arch_local_irq_restore() or when returning
from an interrupt, we can now decide what to do by testing bits in that
field.
We then implement replaying of the missed interrupts either by
re-using the existing exception frame (in exception exit case) or via
the creation of a new one from an assembly trampoline (in the
arch_local_irq_enable case).
This removes the need to play with the decrementer to try to create
fake interrupts, among others.
In addition, this adds a few refinements:
- We no longer hard disable decrementer interrupts that occur
while soft-disabled. We now simply bump the decrementer back to max
(on BookS) or leave it stopped (on BookE) and continue with hard interrupts
enabled, which means that we'll potentially get better sample quality from
performance monitor interrupts.
- Timer, decrementer and doorbell interrupts now hard-enable
shortly after removing the source of the interrupt, which means
they no longer run entirely hard disabled. Again, this will improve
perf sample quality.
- On Book3E 64-bit, we now make the performance monitor interrupt
act as an NMI like Book3S (the necessary C code for that to work
appear to already be present in the FSL perf code, notably calling
nmi_enter instead of irq_enter). (This also fixes a bug where BookE
perfmon interrupts could clobber r14 ... oops)
- We could make "masked" decrementer interrupts act as NMIs when doing
timer-based perf sampling to improve the sample quality.
Signed-off-by-yet: Benjamin Herrenschmidt <benh@kernel.crashing.org>
---
v2:
- Add hard-enable to decrementer, timer and doorbells
- Fix CR clobber in masked irq handling on BookE
- Make embedded perf interrupt act as an NMI
- Add a PACA_HAPPENED_EE_EDGE for use by FSL if they want
to retrigger an interrupt without preventing hard-enable
v3:
- Fix or vs. ori bug on Book3E
- Fix enabling of interrupts for some exceptions on Book3E
v4:
- Fix resend of doorbells on return from interrupt on Book3E
v5:
- Rebased on top of my latest series, which involves some significant
rework of some aspects of the patch.
v6:
- 32-bit compile fix
- more compile fixes with various .config combos
- factor out the asm code to soft-disable interrupts
- remove the C wrapper around preempt_schedule_irq
v7:
- Fix a bug with hard irq state tracking on native power7
2012-03-06 18:27:59 +11:00
/ * Mark i n t e r r u p t s s o f t a n d h a r d d i s a b l e d ( t h e y m i g h t b e e n a b l e d
* in t h e P A C A w h e n d o i n g h o t p l u g )
* /
2017-12-20 09:25:42 +05:30
li r0 ,I R Q S _ D I S A B L E D
2017-12-20 09:25:50 +05:30
stb r0 ,P A C A I R Q S O F T M A S K ( r13 )
powerpc: Rework lazy-interrupt handling
The current implementation of lazy interrupts handling has some
issues that this tries to address.
We don't do the various workarounds we need to do when re-enabling
interrupts in some cases such as when returning from an interrupt
and thus we may still lose or get delayed decrementer or doorbell
interrupts.
The current scheme also makes it much harder to handle the external
"edge" interrupts provided by some BookE processors when using the
EPR facility (External Proxy) and the Freescale Hypervisor.
Additionally, we tend to keep interrupts hard disabled in a number
of cases, such as decrementer interrupts, external interrupts, or
when a masked decrementer interrupt is pending. This is sub-optimal.
This is an attempt at fixing it all in one go by reworking the way
we do the lazy interrupt disabling from the ground up.
The base idea is to replace the "hard_enabled" field with a
"irq_happened" field in which we store a bit mask of what interrupt
occurred while soft-disabled.
When re-enabling, either via arch_local_irq_restore() or when returning
from an interrupt, we can now decide what to do by testing bits in that
field.
We then implement replaying of the missed interrupts either by
re-using the existing exception frame (in exception exit case) or via
the creation of a new one from an assembly trampoline (in the
arch_local_irq_enable case).
This removes the need to play with the decrementer to try to create
fake interrupts, among others.
In addition, this adds a few refinements:
- We no longer hard disable decrementer interrupts that occur
while soft-disabled. We now simply bump the decrementer back to max
(on BookS) or leave it stopped (on BookE) and continue with hard interrupts
enabled, which means that we'll potentially get better sample quality from
performance monitor interrupts.
- Timer, decrementer and doorbell interrupts now hard-enable
shortly after removing the source of the interrupt, which means
they no longer run entirely hard disabled. Again, this will improve
perf sample quality.
- On Book3E 64-bit, we now make the performance monitor interrupt
act as an NMI like Book3S (the necessary C code for that to work
appear to already be present in the FSL perf code, notably calling
nmi_enter instead of irq_enter). (This also fixes a bug where BookE
perfmon interrupts could clobber r14 ... oops)
- We could make "masked" decrementer interrupts act as NMIs when doing
timer-based perf sampling to improve the sample quality.
Signed-off-by-yet: Benjamin Herrenschmidt <benh@kernel.crashing.org>
---
v2:
- Add hard-enable to decrementer, timer and doorbells
- Fix CR clobber in masked irq handling on BookE
- Make embedded perf interrupt act as an NMI
- Add a PACA_HAPPENED_EE_EDGE for use by FSL if they want
to retrigger an interrupt without preventing hard-enable
v3:
- Fix or vs. ori bug on Book3E
- Fix enabling of interrupts for some exceptions on Book3E
v4:
- Fix resend of doorbells on return from interrupt on Book3E
v5:
- Rebased on top of my latest series, which involves some significant
rework of some aspects of the patch.
v6:
- 32-bit compile fix
- more compile fixes with various .config combos
- factor out the asm code to soft-disable interrupts
- remove the C wrapper around preempt_schedule_irq
v7:
- Fix a bug with hard irq state tracking on native power7
2012-03-06 18:27:59 +11:00
li r0 ,P A C A _ I R Q _ H A R D _ D I S
stb r0 ,P A C A I R Q H A P P E N E D ( r13 )
2005-09-26 16:04:21 +10:00
powerpc: Rework lazy-interrupt handling
The current implementation of lazy interrupts handling has some
issues that this tries to address.
We don't do the various workarounds we need to do when re-enabling
interrupts in some cases such as when returning from an interrupt
and thus we may still lose or get delayed decrementer or doorbell
interrupts.
The current scheme also makes it much harder to handle the external
"edge" interrupts provided by some BookE processors when using the
EPR facility (External Proxy) and the Freescale Hypervisor.
Additionally, we tend to keep interrupts hard disabled in a number
of cases, such as decrementer interrupts, external interrupts, or
when a masked decrementer interrupt is pending. This is sub-optimal.
This is an attempt at fixing it all in one go by reworking the way
we do the lazy interrupt disabling from the ground up.
The base idea is to replace the "hard_enabled" field with a
"irq_happened" field in which we store a bit mask of what interrupt
occurred while soft-disabled.
When re-enabling, either via arch_local_irq_restore() or when returning
from an interrupt, we can now decide what to do by testing bits in that
field.
We then implement replaying of the missed interrupts either by
re-using the existing exception frame (in exception exit case) or via
the creation of a new one from an assembly trampoline (in the
arch_local_irq_enable case).
This removes the need to play with the decrementer to try to create
fake interrupts, among others.
In addition, this adds a few refinements:
- We no longer hard disable decrementer interrupts that occur
while soft-disabled. We now simply bump the decrementer back to max
(on BookS) or leave it stopped (on BookE) and continue with hard interrupts
enabled, which means that we'll potentially get better sample quality from
performance monitor interrupts.
- Timer, decrementer and doorbell interrupts now hard-enable
shortly after removing the source of the interrupt, which means
they no longer run entirely hard disabled. Again, this will improve
perf sample quality.
- On Book3E 64-bit, we now make the performance monitor interrupt
act as an NMI like Book3S (the necessary C code for that to work
appear to already be present in the FSL perf code, notably calling
nmi_enter instead of irq_enter). (This also fixes a bug where BookE
perfmon interrupts could clobber r14 ... oops)
- We could make "masked" decrementer interrupts act as NMIs when doing
timer-based perf sampling to improve the sample quality.
Signed-off-by-yet: Benjamin Herrenschmidt <benh@kernel.crashing.org>
---
v2:
- Add hard-enable to decrementer, timer and doorbells
- Fix CR clobber in masked irq handling on BookE
- Make embedded perf interrupt act as an NMI
- Add a PACA_HAPPENED_EE_EDGE for use by FSL if they want
to retrigger an interrupt without preventing hard-enable
v3:
- Fix or vs. ori bug on Book3E
- Fix enabling of interrupts for some exceptions on Book3E
v4:
- Fix resend of doorbells on return from interrupt on Book3E
v5:
- Rebased on top of my latest series, which involves some significant
rework of some aspects of the patch.
v6:
- 32-bit compile fix
- more compile fixes with various .config combos
- factor out the asm code to soft-disable interrupts
- remove the C wrapper around preempt_schedule_irq
v7:
- Fix a bug with hard irq state tracking on native power7
2012-03-06 18:27:59 +11:00
/* Generic kernel entry */
2014-02-04 16:04:35 +11:00
bl s t a r t _ k e r n e l
2005-09-26 16:04:21 +10:00
2006-02-13 18:11:13 +11:00
/* Not reached */
2020-11-30 11:44:04 +11:00
0 : trap
2019-08-26 11:10:23 +00:00
EMIT_ B U G _ E N T R Y 0 b , _ _ F I L E _ _ , _ _ L I N E _ _ , 0
powerpc/64s: Fix section mismatch warnings from boot code
We currently have two section mismatch warnings:
The function __boot_from_prom() references
the function __init prom_init().
The function start_here_common() references
the function __init start_kernel().
The warnings are correct, we do have branches from non-init code into
init code, which is freed after boot. But we don't expect to ever
execute any of that early boot code after boot, if we did that would
be a bug. In particular calling into OF after boot would be fatal
because OF is no longer resident.
So for now fix the warnings by marking the relevant functions as
__REF, which puts them in the ".ref.text" section.
This causes some reordering of the functions in the final link:
@@ -217,10 +217,9 @@
c00000000000b088 t generic_secondary_common_init
c00000000000b124 t __mmu_off
c00000000000b14c t __start_initialization_multiplatform
-c00000000000b1ac t __boot_from_prom
-c00000000000b1ec t __after_prom_start
-c00000000000b260 t p_end
-c00000000000b27c T copy_and_flush
+c00000000000b1ac t __after_prom_start
+c00000000000b220 t p_end
+c00000000000b23c T copy_and_flush
c00000000000b300 T __secondary_start
c00000000000b300 t copy_to_here
c00000000000b344 t start_secondary_prolog
@@ -228,8 +227,9 @@
c00000000000b36c t enable_64b_mode
c00000000000b388 T relative_toc
c00000000000b3a8 t p_toc
-c00000000000b3b0 t start_here_common
-c00000000000b3d0 t start_here_multiplatform
+c00000000000b3b0 t __boot_from_prom
+c00000000000b3f0 t start_here_multiplatform
+c00000000000b480 t start_here_common
c00000000000b880 T system_call_common
c00000000000b974 t system_call
c00000000000b9dc t system_call_exit
In particular __boot_from_prom moves after copy_to_here, which means
it's not copied to zero in the first stage of copy of the kernel to
zero.
But that's OK, because we only call __boot_from_prom before we do the
copy, so it makes no difference when it's copied. The call sequence
is:
__start
-> __start_initialization_multiplatform
-> __boot_from_prom
-> __start
-> __start_initialization_multiplatform
-> __after_prom_start
-> copy_and_flush
-> copy_and_flush (relocated to 0)
-> start_here_multiplatform
-> early_setup
Reported-by: Mauricio Faria de Oliveira <mauricfo@linux.ibm.com>
Reported-by: Roman Bolshakov <r.bolshakov@yadro.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20200225031328.14676-1-mpe@ellerman.id.au
2020-02-25 14:13:28 +11:00
.previous
