From 89067c2daf3d9e0ce51c768589e79e845e6fda42 Mon Sep 17 00:00:00 2001 From: Chris Metcalf Date: Mon, 16 Mar 2015 15:04:05 -0400 Subject: [PATCH 1/8] tile: use si_int instead of si_ptr for compat_siginfo To be compatible with the generic get_compat_sigevent(), the copy_siginfo_to_user32() and thus copy_siginfo_from_user32() have to use si_int instead of si_ptr. Using si_ptr means that for the case of ILP32 compat code running in big-endian mode, we would end up copying the high 32 bits of the pointer value into si_int instead of the desired low 32 bits. Signed-off-by: Chris Metcalf Cc: Catalin Marinas --- arch/tile/kernel/compat_signal.c | 11 +++++------ 1 file changed, 5 insertions(+), 6 deletions(-) diff --git a/arch/tile/kernel/compat_signal.c b/arch/tile/kernel/compat_signal.c index 8c5abf2e4794..c52e6fd3c438 100644 --- a/arch/tile/kernel/compat_signal.c +++ b/arch/tile/kernel/compat_signal.c @@ -68,7 +68,7 @@ int copy_siginfo_to_user32(struct compat_siginfo __user *to, const siginfo_t *fr if (from->si_code < 0) { err |= __put_user(from->si_pid, &to->si_pid); err |= __put_user(from->si_uid, &to->si_uid); - err |= __put_user(ptr_to_compat(from->si_ptr), &to->si_ptr); + err |= __put_user(from->si_int, &to->si_int); } else { /* * First 32bits of unions are always present: @@ -93,8 +93,7 @@ int copy_siginfo_to_user32(struct compat_siginfo __user *to, const siginfo_t *fr break; case __SI_TIMER >> 16: err |= __put_user(from->si_overrun, &to->si_overrun); - err |= __put_user(ptr_to_compat(from->si_ptr), - &to->si_ptr); + err |= __put_user(from->si_int, &to->si_int); break; /* This is not generated by the kernel as of now. */ case __SI_RT >> 16: @@ -110,19 +109,19 @@ int copy_siginfo_to_user32(struct compat_siginfo __user *to, const siginfo_t *fr int copy_siginfo_from_user32(siginfo_t *to, struct compat_siginfo __user *from) { int err; - u32 ptr32; if (!access_ok(VERIFY_READ, from, sizeof(struct compat_siginfo))) return -EFAULT; + memset(to, 0, sizeof(*to)); + err = __get_user(to->si_signo, &from->si_signo); err |= __get_user(to->si_errno, &from->si_errno); err |= __get_user(to->si_code, &from->si_code); err |= __get_user(to->si_pid, &from->si_pid); err |= __get_user(to->si_uid, &from->si_uid); - err |= __get_user(ptr32, &from->si_ptr); - to->si_ptr = compat_ptr(ptr32); + err |= __get_user(to->si_int, &from->si_int); return err; } From 5a3b4e8000c1ce476f5e8babd62c580457561f34 Mon Sep 17 00:00:00 2001 From: Davidlohr Bueso Date: Wed, 25 Feb 2015 13:58:35 -0800 Subject: [PATCH 2/8] tile/elf: reorganize notify_exec() In the future mm->exe_file will be done without mmap_sem serialization, thus isolate and reorganize the tile elf code to make the transition easier. Good users will, make use of the more standard get_mm_exe_file(), requiring only holding the mmap_sem to read the value, and relying on reference counting to make sure that the exe file won't dissappear underneath us. The visible effects of this patch are: o We now take and drop the mmap_sem more often. Instead of just in arch_setup_additional_pages(), we also do it in: 1) get_mm_exe_file() 2) to get the mm->vm_file and notify the simulator. [Note that 1) will disappear once we change the locking rules for exe_file.] o We avoid getting a free page and doing d_path() while holding the mmap_sem. This requires reordering the checks. Signed-off-by: Davidlohr Bueso Signed-off-by: Chris Metcalf --- arch/tile/mm/elf.c | 47 ++++++++++++++++++++++++++++------------------ 1 file changed, 29 insertions(+), 18 deletions(-) diff --git a/arch/tile/mm/elf.c b/arch/tile/mm/elf.c index 23f044e8a7ab..f7ddae3725a4 100644 --- a/arch/tile/mm/elf.c +++ b/arch/tile/mm/elf.c @@ -17,6 +17,7 @@ #include #include #include +#include #include #include #include @@ -39,30 +40,34 @@ static void sim_notify_exec(const char *binary_name) static int notify_exec(struct mm_struct *mm) { + int ret = 0; char *buf, *path; struct vm_area_struct *vma; + struct file *exe_file; if (!sim_is_simulator()) return 1; - if (mm->exe_file == NULL) - return 0; - - for (vma = current->mm->mmap; ; vma = vma->vm_next) { - if (vma == NULL) - return 0; - if (vma->vm_file == mm->exe_file) - break; - } - buf = (char *) __get_free_page(GFP_KERNEL); if (buf == NULL) return 0; - path = d_path(&mm->exe_file->f_path, buf, PAGE_SIZE); - if (IS_ERR(path)) { - free_page((unsigned long)buf); - return 0; + exe_file = get_mm_exe_file(mm); + if (exe_file == NULL) + goto done_free; + + path = d_path(&exe_file->f_path, buf, PAGE_SIZE); + if (IS_ERR(path)) + goto done_put; + + down_read(&mm->mmap_sem); + for (vma = current->mm->mmap; ; vma = vma->vm_next) { + if (vma == NULL) { + up_read(&mm->mmap_sem); + goto done_put; + } + if (vma->vm_file == exe_file) + break; } /* @@ -80,14 +85,20 @@ static int notify_exec(struct mm_struct *mm) __insn_mtspr(SPR_SIM_CONTROL, (SIM_CONTROL_DLOPEN | (c << _SIM_CONTROL_OPERATOR_BITS))); - if (c == '\0') + if (c == '\0') { + ret = 1; /* success */ break; + } } } + up_read(&mm->mmap_sem); sim_notify_exec(path); +done_put: + fput(exe_file); +done_free: free_page((unsigned long)buf); - return 1; + return ret; } /* Notify a running simulator, if any, that we loaded an interpreter. */ @@ -109,8 +120,6 @@ int arch_setup_additional_pages(struct linux_binprm *bprm, struct mm_struct *mm = current->mm; int retval = 0; - down_write(&mm->mmap_sem); - /* * Notify the simulator that an exec just occurred. * If we can't find the filename of the mapping, just use @@ -119,6 +128,8 @@ int arch_setup_additional_pages(struct linux_binprm *bprm, if (!notify_exec(mm)) sim_notify_exec(bprm->filename); + down_write(&mm->mmap_sem); + retval = setup_vdso_pages(); #ifndef __tilegx__ From 9088616fb3c4e0d3ec4efb20378691066a218f9c Mon Sep 17 00:00:00 2001 From: Colin Ian King Date: Mon, 16 Mar 2015 16:14:02 -0400 Subject: [PATCH 3/8] arch: tile: fix null pointer dereference on pt_regs pointer Cppcheck reports the following issue: [arch/tile/kernel/stack.c:116]: (error) Possible null pointer dereference: p In this case, on reporting on an odd fault, p is set to NULL and immediately afterwords p is dereferenced iff !kbt->profile is false. Rather than doing this check just return NULL rather than falling through to the potential null pointer dereference (since the original intentional outcome would be to return NULL anyhow) for this odd fault case. Signed-off-by: Colin Ian King Signed-off-by: Chris Metcalf [tweaked lightly] --- arch/tile/kernel/stack.c | 15 ++++++++------- 1 file changed, 8 insertions(+), 7 deletions(-) diff --git a/arch/tile/kernel/stack.c b/arch/tile/kernel/stack.c index 7ff5afdbd3aa..c42dce50acd8 100644 --- a/arch/tile/kernel/stack.c +++ b/arch/tile/kernel/stack.c @@ -108,14 +108,15 @@ static struct pt_regs *valid_fault_handler(struct KBacktraceIterator* kbt) p->sp < PAGE_OFFSET && p->sp != 0) { if (kbt->verbose) pr_err(" <%s while in user mode>\n", fault); - } else if (kbt->verbose) { - pr_err(" (odd fault: pc %#lx, sp %#lx, ex1 %#lx?)\n", - p->pc, p->sp, p->ex1); - p = NULL; + } else { + if (kbt->verbose) + pr_err(" (odd fault: pc %#lx, sp %#lx, ex1 %#lx?)\n", + p->pc, p->sp, p->ex1); + return NULL; } - if (!kbt->profile || ((1ULL << p->faultnum) & QUEUED_INTERRUPTS) == 0) - return p; - return NULL; + if (kbt->profile && ((1ULL << p->faultnum) & QUEUED_INTERRUPTS) != 0) + return NULL; + return p; } /* Is the pc pointing to a sigreturn trampoline? */ From b340c656af6317e28b466996a72cca019d97b42d Mon Sep 17 00:00:00 2001 From: Chris Metcalf Date: Mon, 23 Mar 2015 11:21:23 -0400 Subject: [PATCH 4/8] tile: support arch_irq_work_raise Tile includes a hypervisor hook to deliver messages to arbitrary tiles, so we can use that to raise an interrupt as soon as possible on our own core. Unfortunately the Tilera hypervisor disabled that support on principle in previous releases, but it will be available in MDE 4.3.4 and later. Signed-off-by: Chris Metcalf Acked-by: Frederic Weisbecker --- arch/tile/include/asm/Kbuild | 1 - arch/tile/include/asm/irq_work.h | 14 ++++++++++++++ arch/tile/include/asm/smp.h | 1 + arch/tile/kernel/smp.c | 32 +++++++++++++++++++++++++++++++- 4 files changed, 46 insertions(+), 2 deletions(-) create mode 100644 arch/tile/include/asm/irq_work.h diff --git a/arch/tile/include/asm/Kbuild b/arch/tile/include/asm/Kbuild index b4c488b65745..f5433e0e34e0 100644 --- a/arch/tile/include/asm/Kbuild +++ b/arch/tile/include/asm/Kbuild @@ -16,7 +16,6 @@ generic-y += ioctl.h generic-y += ioctls.h generic-y += ipcbuf.h generic-y += irq_regs.h -generic-y += irq_work.h generic-y += local.h generic-y += local64.h generic-y += mcs_spinlock.h diff --git a/arch/tile/include/asm/irq_work.h b/arch/tile/include/asm/irq_work.h new file mode 100644 index 000000000000..48af33a61a2c --- /dev/null +++ b/arch/tile/include/asm/irq_work.h @@ -0,0 +1,14 @@ +#ifndef __ASM_IRQ_WORK_H +#define __ASM_IRQ_WORK_H + +static inline bool arch_irq_work_has_interrupt(void) +{ +#ifdef CONFIG_SMP + extern bool self_interrupt_ok; + return self_interrupt_ok; +#else + return false; +#endif +} + +#endif /* __ASM_IRQ_WORK_H */ diff --git a/arch/tile/include/asm/smp.h b/arch/tile/include/asm/smp.h index 9a326b64f7ae..735e7f144733 100644 --- a/arch/tile/include/asm/smp.h +++ b/arch/tile/include/asm/smp.h @@ -69,6 +69,7 @@ static inline int xy_to_cpu(int x, int y) #define MSG_TAG_STOP_CPU 2 #define MSG_TAG_CALL_FUNCTION_MANY 3 #define MSG_TAG_CALL_FUNCTION_SINGLE 4 +#define MSG_TAG_IRQ_WORK 5 /* Hook for the generic smp_call_function_many() routine. */ static inline void arch_send_call_function_ipi_mask(struct cpumask *mask) diff --git a/arch/tile/kernel/smp.c b/arch/tile/kernel/smp.c index d3c4ed780ce2..07e3ff5cc740 100644 --- a/arch/tile/kernel/smp.c +++ b/arch/tile/kernel/smp.c @@ -18,6 +18,7 @@ #include #include #include +#include #include #include #include @@ -33,6 +34,8 @@ EXPORT_SYMBOL(smp_topology); static unsigned long __iomem *ipi_mappings[NR_CPUS]; #endif +/* Does messaging work correctly to the local cpu? */ +bool self_interrupt_ok; /* * Top-level send_IPI*() functions to send messages to other cpus. @@ -147,6 +150,10 @@ void evaluate_message(int tag) generic_smp_call_function_single_interrupt(); break; + case MSG_TAG_IRQ_WORK: /* Invoke IRQ work */ + irq_work_run(); + break; + default: panic("Unknown IPI message tag %d", tag); break; @@ -186,6 +193,15 @@ void flush_icache_range(unsigned long start, unsigned long end) EXPORT_SYMBOL(flush_icache_range); +#ifdef CONFIG_IRQ_WORK +void arch_irq_work_raise(void) +{ + if (arch_irq_work_has_interrupt()) + send_IPI_single(smp_processor_id(), MSG_TAG_IRQ_WORK); +} +#endif + + /* Called when smp_send_reschedule() triggers IRQ_RESCHEDULE. */ static irqreturn_t handle_reschedule_ipi(int irq, void *token) { @@ -203,8 +219,22 @@ static struct irqaction resched_action = { void __init ipi_init(void) { + int cpu = smp_processor_id(); + HV_Recipient recip = { .y = cpu_y(cpu), .x = cpu_x(cpu), + .state = HV_TO_BE_SENT }; + int tag = MSG_TAG_CALL_FUNCTION_SINGLE; + + /* + * Test if we can message ourselves for arch_irq_work_raise. + * This functionality is only available in the Tilera hypervisor + * in versions 4.3.4 and following. + */ + if (hv_send_message(&recip, 1, (HV_VirtAddr)&tag, sizeof(tag)) == 1) + self_interrupt_ok = true; + else + pr_warn("Older hypervisor: disabling fast irq_work_raise\n"); + #if CHIP_HAS_IPI() - int cpu; /* Map IPI trigger MMIO addresses. */ for_each_possible_cpu(cpu) { HV_Coord tile; From 49e4e15619cd7cd9fc275d460fae2a95c1337fcc Mon Sep 17 00:00:00 2001 From: Chris Metcalf Date: Mon, 23 Mar 2015 14:23:58 -0400 Subject: [PATCH 5/8] tile: support CONTEXT_TRACKING and thus NOHZ_FULL Add the TIF_NOHZ flag appropriately. Add call to user_exit() on entry to do_work_pending() and on entry to syscalls via do_syscall_trace_enter(), and also the top of do_syscall_trace_exit() just because it's done in x86. Add call to user_enter() at the bottom of do_work_pending() once we have no more work to do before returning to userspace. Wrap all the trap code in exception_enter() / exception_exit(). Signed-off-by: Chris Metcalf Acked-by: Frederic Weisbecker --- arch/tile/Kconfig | 1 + arch/tile/include/asm/thread_info.h | 9 ++++++--- arch/tile/kernel/process.c | 12 ++++++++---- arch/tile/kernel/ptrace.c | 22 ++++++++++++++++++++-- arch/tile/kernel/single_step.c | 3 +++ arch/tile/kernel/traps.c | 16 +++++++++------- arch/tile/kernel/unaligned.c | 22 +++++++++++++--------- arch/tile/mm/fault.c | 10 +++++++--- 8 files changed, 67 insertions(+), 28 deletions(-) diff --git a/arch/tile/Kconfig b/arch/tile/Kconfig index 7cca41842a9e..c3a31f8bb09c 100644 --- a/arch/tile/Kconfig +++ b/arch/tile/Kconfig @@ -27,6 +27,7 @@ config TILE select ARCH_HAS_ATOMIC64_DEC_IF_POSITIVE select HAVE_DEBUG_STACKOVERFLOW select ARCH_WANT_FRAME_POINTERS + select HAVE_CONTEXT_TRACKING # FIXME: investigate whether we need/want these options. # select HAVE_IOREMAP_PROT diff --git a/arch/tile/include/asm/thread_info.h b/arch/tile/include/asm/thread_info.h index 96c14c1430d8..6130a3db505b 100644 --- a/arch/tile/include/asm/thread_info.h +++ b/arch/tile/include/asm/thread_info.h @@ -126,6 +126,7 @@ extern void _cpu_idle(void); #define TIF_NOTIFY_RESUME 8 /* callback before returning to user */ #define TIF_SYSCALL_TRACEPOINT 9 /* syscall tracepoint instrumentation */ #define TIF_POLLING_NRFLAG 10 /* idle is polling for TIF_NEED_RESCHED */ +#define TIF_NOHZ 11 /* in adaptive nohz mode */ #define _TIF_SIGPENDING (1< #include #include +#include #include #include #include @@ -474,6 +475,8 @@ int do_work_pending(struct pt_regs *regs, u32 thread_info_flags) if (!user_mode(regs)) return 0; + user_exit(); + /* Enable interrupts; they are disabled again on return to caller. */ local_irq_enable(); @@ -496,11 +499,12 @@ int do_work_pending(struct pt_regs *regs, u32 thread_info_flags) tracehook_notify_resume(regs); return 1; } - if (thread_info_flags & _TIF_SINGLESTEP) { + if (thread_info_flags & _TIF_SINGLESTEP) single_step_once(regs); - return 0; - } - panic("work_pending: bad flags %#x\n", thread_info_flags); + + user_enter(); + + return 0; } unsigned long get_wchan(struct task_struct *p) diff --git a/arch/tile/kernel/ptrace.c b/arch/tile/kernel/ptrace.c index de98c6ddf136..f84eed8243da 100644 --- a/arch/tile/kernel/ptrace.c +++ b/arch/tile/kernel/ptrace.c @@ -22,6 +22,7 @@ #include #include #include +#include #include #include @@ -252,12 +253,21 @@ long compat_arch_ptrace(struct task_struct *child, compat_long_t request, int do_syscall_trace_enter(struct pt_regs *regs) { - if (test_thread_flag(TIF_SYSCALL_TRACE)) { + u32 work = ACCESS_ONCE(current_thread_info()->flags); + + /* + * If TIF_NOHZ is set, we are required to call user_exit() before + * doing anything that could touch RCU. + */ + if (work & _TIF_NOHZ) + user_exit(); + + if (work & _TIF_SYSCALL_TRACE) { if (tracehook_report_syscall_entry(regs)) regs->regs[TREG_SYSCALL_NR] = -1; } - if (test_thread_flag(TIF_SYSCALL_TRACEPOINT)) + if (work & _TIF_SYSCALL_TRACEPOINT) trace_sys_enter(regs, regs->regs[TREG_SYSCALL_NR]); return regs->regs[TREG_SYSCALL_NR]; @@ -267,6 +277,12 @@ void do_syscall_trace_exit(struct pt_regs *regs) { long errno; + /* + * We may come here right after calling schedule_user() + * in which case we can be in RCU user mode. + */ + user_exit(); + /* * The standard tile calling convention returns the value (or negative * errno) in r0, and zero (or positive errno) in r1. @@ -303,5 +319,7 @@ void send_sigtrap(struct task_struct *tsk, struct pt_regs *regs) /* Handle synthetic interrupt delivered only by the simulator. */ void __kprobes do_breakpoint(struct pt_regs* regs, int fault_num) { + enum ctx_state prev_state = exception_enter(); send_sigtrap(current, regs); + exception_exit(prev_state); } diff --git a/arch/tile/kernel/single_step.c b/arch/tile/kernel/single_step.c index 862973074bf9..53f7b9def07b 100644 --- a/arch/tile/kernel/single_step.c +++ b/arch/tile/kernel/single_step.c @@ -23,6 +23,7 @@ #include #include #include +#include #include #include #include @@ -738,6 +739,7 @@ static DEFINE_PER_CPU(unsigned long, ss_saved_pc); void gx_singlestep_handle(struct pt_regs *regs, int fault_num) { + enum ctx_state prev_state = exception_enter(); unsigned long *ss_pc = this_cpu_ptr(&ss_saved_pc); struct thread_info *info = (void *)current_thread_info(); int is_single_step = test_ti_thread_flag(info, TIF_SINGLESTEP); @@ -754,6 +756,7 @@ void gx_singlestep_handle(struct pt_regs *regs, int fault_num) __insn_mtspr(SPR_SINGLE_STEP_CONTROL_K, control); send_sigtrap(current, regs); } + exception_exit(prev_state); } diff --git a/arch/tile/kernel/traps.c b/arch/tile/kernel/traps.c index bf841ca517bb..312fc134c1cb 100644 --- a/arch/tile/kernel/traps.c +++ b/arch/tile/kernel/traps.c @@ -20,6 +20,7 @@ #include #include #include +#include #include #include #include @@ -253,6 +254,7 @@ static int do_bpt(struct pt_regs *regs) void __kprobes do_trap(struct pt_regs *regs, int fault_num, unsigned long reason) { + enum ctx_state prev_state = exception_enter(); siginfo_t info = { 0 }; int signo, code; unsigned long address = 0; @@ -261,7 +263,7 @@ void __kprobes do_trap(struct pt_regs *regs, int fault_num, /* Handle breakpoints, etc. */ if (is_kernel && fault_num == INT_ILL && do_bpt(regs)) - return; + goto done; /* Re-enable interrupts, if they were previously enabled. */ if (!(regs->flags & PT_FLAGS_DISABLE_IRQ)) @@ -275,7 +277,7 @@ void __kprobes do_trap(struct pt_regs *regs, int fault_num, const char *name; char buf[100]; if (fixup_exception(regs)) /* ILL_TRANS or UNALIGN_DATA */ - return; + goto done; if (fault_num >= 0 && fault_num < ARRAY_SIZE(int_name) && int_name[fault_num] != NULL) @@ -294,7 +296,6 @@ void __kprobes do_trap(struct pt_regs *regs, int fault_num, fault_num, name, regs->pc, buf); show_regs(regs); do_exit(SIGKILL); /* FIXME: implement i386 die() */ - return; } switch (fault_num) { @@ -308,7 +309,6 @@ void __kprobes do_trap(struct pt_regs *regs, int fault_num, pr_err("Unreadable instruction for INT_ILL: %#lx\n", regs->pc); do_exit(SIGKILL); - return; } if (!special_ill(instr, &signo, &code)) { signo = SIGILL; @@ -319,7 +319,7 @@ void __kprobes do_trap(struct pt_regs *regs, int fault_num, case INT_GPV: #if CHIP_HAS_TILE_DMA() if (retry_gpv(reason)) - return; + goto done; #endif /*FALLTHROUGH*/ case INT_UDN_ACCESS: @@ -346,7 +346,7 @@ void __kprobes do_trap(struct pt_regs *regs, int fault_num, if (!state || (void __user *)(regs->pc) != state->buffer) { single_step_once(regs); - return; + goto done; } } #endif @@ -380,7 +380,6 @@ void __kprobes do_trap(struct pt_regs *regs, int fault_num, #endif default: panic("Unexpected do_trap interrupt number %d", fault_num); - return; } info.si_signo = signo; @@ -391,6 +390,9 @@ void __kprobes do_trap(struct pt_regs *regs, int fault_num, if (signo != SIGTRAP) trace_unhandled_signal("trap", regs, address, signo); force_sig_info(signo, &info, current); + +done: + exception_exit(prev_state); } void kernel_double_fault(int dummy, ulong pc, ulong lr, ulong sp, ulong r52) diff --git a/arch/tile/kernel/unaligned.c b/arch/tile/kernel/unaligned.c index 7d9a83be0aca..d075f92ccee0 100644 --- a/arch/tile/kernel/unaligned.c +++ b/arch/tile/kernel/unaligned.c @@ -25,6 +25,7 @@ #include #include #include +#include #include #include #include @@ -1448,6 +1449,7 @@ void jit_bundle_gen(struct pt_regs *regs, tilegx_bundle_bits bundle, void do_unaligned(struct pt_regs *regs, int vecnum) { + enum ctx_state prev_state = exception_enter(); tilegx_bundle_bits __user *pc; tilegx_bundle_bits bundle; struct thread_info *info = current_thread_info(); @@ -1487,12 +1489,11 @@ void do_unaligned(struct pt_regs *regs, int vecnum) (int)unaligned_fixup, (unsigned long long)regs->ex1, (unsigned long long)regs->pc); - return; + } else { + /* Not fixable. Go panic. */ + panic("Unalign exception in Kernel. pc=%lx", + regs->pc); } - /* Not fixable. Go panic. */ - panic("Unalign exception in Kernel. pc=%lx", - regs->pc); - return; } else { /* * Try to fix the exception. If we can't, panic the @@ -1501,8 +1502,8 @@ void do_unaligned(struct pt_regs *regs, int vecnum) bundle = GX_INSN_BSWAP( *((tilegx_bundle_bits *)(regs->pc))); jit_bundle_gen(regs, bundle, align_ctl); - return; } + goto done; } /* @@ -1526,7 +1527,7 @@ void do_unaligned(struct pt_regs *regs, int vecnum) trace_unhandled_signal("unaligned fixup trap", regs, 0, SIGBUS); force_sig_info(info.si_signo, &info, current); - return; + goto done; } @@ -1543,7 +1544,7 @@ void do_unaligned(struct pt_regs *regs, int vecnum) trace_unhandled_signal("segfault in unalign fixup", regs, (unsigned long)info.si_addr, SIGSEGV); force_sig_info(info.si_signo, &info, current); - return; + goto done; } if (!info->unalign_jit_base) { @@ -1578,7 +1579,7 @@ void do_unaligned(struct pt_regs *regs, int vecnum) if (IS_ERR((void __force *)user_page)) { pr_err("Out of kernel pages trying do_mmap\n"); - return; + goto done; } /* Save the address in the thread_info struct */ @@ -1591,6 +1592,9 @@ void do_unaligned(struct pt_regs *regs, int vecnum) /* Generate unalign JIT */ jit_bundle_gen(regs, GX_INSN_BSWAP(bundle), align_ctl); + +done: + exception_exit(prev_state); } #endif /* __tilegx__ */ diff --git a/arch/tile/mm/fault.c b/arch/tile/mm/fault.c index 0f61a73534e6..e83cc999da02 100644 --- a/arch/tile/mm/fault.c +++ b/arch/tile/mm/fault.c @@ -35,6 +35,7 @@ #include #include #include +#include #include #include @@ -702,6 +703,7 @@ void do_page_fault(struct pt_regs *regs, int fault_num, unsigned long address, unsigned long write) { int is_page_fault; + enum ctx_state prev_state = exception_enter(); #ifdef CONFIG_KPROBES /* @@ -711,7 +713,7 @@ void do_page_fault(struct pt_regs *regs, int fault_num, */ if (notify_die(DIE_PAGE_FAULT, "page fault", regs, -1, regs->faultnum, SIGSEGV) == NOTIFY_STOP) - return; + goto done; #endif #ifdef __tilegx__ @@ -750,7 +752,6 @@ void do_page_fault(struct pt_regs *regs, int fault_num, current->comm, current->pid, pc, address); show_regs(regs); do_group_exit(SIGKILL); - return; } } #else @@ -834,12 +835,15 @@ void do_page_fault(struct pt_regs *regs, int fault_num, async->is_fault = is_page_fault; async->is_write = write; async->address = address; - return; + goto done; } } #endif handle_page_fault(regs, fault_num, is_page_fault, address, write); + +done: + exception_exit(prev_state); } From a84f24230c137a4e0ab14185e9175798ca1b0376 Mon Sep 17 00:00:00 2001 From: Chris Metcalf Date: Fri, 27 Mar 2015 14:35:31 -0400 Subject: [PATCH 6/8] tile: map data region shadow of kernel as R/W This is necessary for things like reading /proc/kcore, doing ftrace, etc. It happens by default when using huge pages to map the kernel data, but not when using small pages. Signed-off-by: Chris Metcalf --- arch/tile/mm/init.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) diff --git a/arch/tile/mm/init.c b/arch/tile/mm/init.c index ace32d7d3864..5bd252e3fdc5 100644 --- a/arch/tile/mm/init.c +++ b/arch/tile/mm/init.c @@ -233,9 +233,12 @@ static pgprot_t __init init_pgprot(ulong address) if (kdata_huge) return construct_pgprot(PAGE_KERNEL, PAGE_HOME_HASH); - /* We map the aliased pages of permanent text inaccessible. */ + /* + * We map the aliased pages of permanent text so we can + * update them if necessary, for ftrace, etc. + */ if (address < (ulong) _sinittext - CODE_DELTA) - return PAGE_NONE; + return construct_pgprot(PAGE_KERNEL, PAGE_HOME_HASH); /* We map read-only data non-coherent for performance. */ if ((address >= (ulong) __start_rodata && From 437d3e124d25daaa671bfecfd4015ecd2503a955 Mon Sep 17 00:00:00 2001 From: Tony Lu Date: Fri, 27 Mar 2015 14:46:38 -0400 Subject: [PATCH 7/8] tile: ftrace: fix function_graph tracer issues - Add support for ARCH_SUPPORTS_FTRACE_OPS - Replace the instruction in ftrace_call with the bundle {move r10, lr; jal ftrace_stub}, so that the lr contains the right value after returning from ftrace_stub. An alternative fix might be to leave the instruction in ftrace_call alone when it is being updated with ftrace_stub. Signed-off-by: Tony Lu Signed-off-by: Chris Metcalf --- arch/tile/include/asm/ftrace.h | 2 ++ arch/tile/kernel/ftrace.c | 6 +++++- arch/tile/kernel/mcount_64.S | 7 ++++++- 3 files changed, 13 insertions(+), 2 deletions(-) diff --git a/arch/tile/include/asm/ftrace.h b/arch/tile/include/asm/ftrace.h index 13a9bb81a8ab..738d239b792f 100644 --- a/arch/tile/include/asm/ftrace.h +++ b/arch/tile/include/asm/ftrace.h @@ -23,6 +23,8 @@ #ifndef __ASSEMBLY__ extern void __mcount(void); +#define ARCH_SUPPORTS_FTRACE_OPS 1 + #ifdef CONFIG_DYNAMIC_FTRACE static inline unsigned long ftrace_call_adjust(unsigned long addr) { diff --git a/arch/tile/kernel/ftrace.c b/arch/tile/kernel/ftrace.c index 8d52d83cc516..0c0996175b1e 100644 --- a/arch/tile/kernel/ftrace.c +++ b/arch/tile/kernel/ftrace.c @@ -74,7 +74,11 @@ static unsigned long ftrace_gen_branch(unsigned long pc, unsigned long addr, create_JumpOff_X1(pcrel_by_instr); } - if (addr == FTRACE_ADDR) { + /* + * Also put { move r10, lr; jal ftrace_stub } in a bundle, which + * is used to replace the instruction in address ftrace_call. + */ + if (addr == FTRACE_ADDR || addr == (unsigned long)ftrace_stub) { /* opcode: or r10, lr, zero */ opcode_x0 = create_Dest_X0(10) | diff --git a/arch/tile/kernel/mcount_64.S b/arch/tile/kernel/mcount_64.S index 3c2b8d5e1d1a..6c6702451962 100644 --- a/arch/tile/kernel/mcount_64.S +++ b/arch/tile/kernel/mcount_64.S @@ -81,7 +81,12 @@ STD_ENTRY(ftrace_caller) /* arg1: self return address */ /* arg2: parent's return address */ - { move r0, lr; move r1, r10 } + /* arg3: ftrace_ops */ + /* arg4: regs (but make it NULL) */ + { move r0, lr; moveli r2, hw2_last(function_trace_op) } + { move r1, r10; shl16insli r2, r2, hw1(function_trace_op) } + { movei r3, 0; shl16insli r2, r2, hw0(function_trace_op) } + ld r2,r2 .global ftrace_call ftrace_call: From 128f3cb9398b5eeb4ee04b60bd5e314f5c122821 Mon Sep 17 00:00:00 2001 From: Chris Metcalf Date: Mon, 30 Mar 2015 16:33:00 -0400 Subject: [PATCH 8/8] tile: nohz: warn if nohz_full uses hypervisor shared cores The "hypervisor shared" cores are ones that the Tilera hypervisor uses to receive interrupts to manage hypervisor-owned devices. It's a bad idea to try to use those cores with nohz_full, since they will get interrupted unpredictably -- and invisibly to Linux tracing tools, since the interrupts are delivered at a higher privilege level to the Tilera hypervisor. Generate a clear warning at boot up that this doesn't end well for the nohz_full cores in question. Signed-off-by: Chris Metcalf --- arch/tile/include/hv/hypervisor.h | 6 +++++- arch/tile/kernel/setup.c | 23 +++++++++++++++++++++++ 2 files changed, 28 insertions(+), 1 deletion(-) diff --git a/arch/tile/include/hv/hypervisor.h b/arch/tile/include/hv/hypervisor.h index dfcdeb61ba34..e0e6af4e783b 100644 --- a/arch/tile/include/hv/hypervisor.h +++ b/arch/tile/include/hv/hypervisor.h @@ -961,7 +961,11 @@ typedef enum { HV_INQ_TILES_HFH_CACHE = 2, /** The set of tiles that can be legally used as a LOTAR for a PTE. */ - HV_INQ_TILES_LOTAR = 3 + HV_INQ_TILES_LOTAR = 3, + + /** The set of "shared" driver tiles that the hypervisor may + * periodically interrupt. */ + HV_INQ_TILES_SHARED = 4 } HV_InqTileSet; /** Returns specific information about various sets of tiles within the diff --git a/arch/tile/kernel/setup.c b/arch/tile/kernel/setup.c index f1f579914952..7833b2ccdfbc 100644 --- a/arch/tile/kernel/setup.c +++ b/arch/tile/kernel/setup.c @@ -32,6 +32,7 @@ #include #include #include +#include #include #include #include @@ -1390,6 +1391,28 @@ static int __init dataplane(char *str) early_param("dataplane", dataplane); +#ifdef CONFIG_NO_HZ_FULL +/* Warn if hypervisor shared cpus are marked as nohz_full. */ +static int __init check_nohz_full_cpus(void) +{ + struct cpumask shared; + int cpu; + + if (hv_inquire_tiles(HV_INQ_TILES_SHARED, + (HV_VirtAddr) shared.bits, sizeof(shared)) < 0) { + pr_warn("WARNING: No support for inquiring hv shared tiles\n"); + return 0; + } + for_each_cpu(cpu, &shared) { + if (tick_nohz_full_cpu(cpu)) + pr_warn("WARNING: nohz_full cpu %d receives hypervisor interrupts!\n", + cpu); + } + return 0; +} +arch_initcall(check_nohz_full_cpus); +#endif + #ifdef CONFIG_CMDLINE_BOOL static char __initdata builtin_cmdline[COMMAND_LINE_SIZE] = CONFIG_CMDLINE; #endif