IF YOU WOULD LIKE TO GET AN ACCOUNT, please write an
email to Administrator. User accounts are meant only to access repo
and report issues and/or generate pull requests.
This is a purpose-specific Git hosting for
BaseALT
projects. Thank you for your understanding!
Только зарегистрированные пользователи имеют доступ к сервису!
Для получения аккаунта, обратитесь к администратору.
Problem: In glusterfs code base we call mutex_lock/unlock to take
reference/dereference for a object.Sometime it could be
reason for lock contention also.
Solution: There is no need to use mutex to increase/decrease ref
counter, instead of using mutex use gcc builtin ATOMIC
operation.
Test: I have not observed yet how much performance gain after apply
this patch specific to glusterfs but i have tested same
with below small program(mutex and atomic both) and
get good difference.
static int numOuterLoops;
static void *
threadFunc(void *arg)
{
int j;
for (j = 0; j < numOuterLoops; j++) {
__atomic_add_fetch (&glob, 1,__ATOMIC_ACQ_REL);
}
return NULL;
}
int
main(int argc, char *argv[])
{
int opt, s, j;
int numThreads;
pthread_t *thread;
int verbose;
int64_t n = 0;
if (argc < 2 ) {
printf(" Please provide 2 args Num of threads && Outer Loop\n");
exit (-1);
}
numThreads = atoi(argv[1]);
numOuterLoops = atoi (argv[2]);
if (1) {
printf("\tthreads: %d; outer loops: %d;\n",
numThreads, numOuterLoops);
}
thread = calloc(numThreads, sizeof(pthread_t));
if (thread == NULL) {
printf ("calloc error so exit\n");
exit (-1);
}
__atomic_store (&glob, &n, __ATOMIC_RELEASE);
for (j = 0; j < numThreads; j++) {
s = pthread_create(&thread[j], NULL, threadFunc, NULL);
if (s != 0) {
printf ("pthread_create failed so exit\n");
exit (-1);
}
}
for (j = 0; j < numThreads; j++) {
s = pthread_join(thread[j], NULL);
if (s != 0) {
printf ("pthread_join failed so exit\n");
exit (-1);
}
}
printf("glob value is %ld\n",__atomic_load_n (&glob,__ATOMIC_RELAXED));
exit(0);
}
time ./thr_count 800 800000
threads: 800; outer loops: 800000;
glob value is 640000000
real 1m10.288s
user 0m57.269s
sys 3m31.565s
time ./thr_count_atomic 800 800000
threads: 800; outer loops: 800000;
glob value is 640000000
real 0m20.313s
user 1m20.558s
sys 0m0.028
Change-Id: Ie5030a52ea264875e002e108dd4b207b15ab7cc7
Signed-off-by: Mohit Agrawal <moagrawa@redhat.com>
With configure --enable-debug, add all object allocations
to a list in the corresponding mem_acct_rec. This
allows us to see all objects of a particular type
and allows for additional debugging in case of memory
leaks.
This is not compiled in by default and must be explicitly
enabled. It is intended to be used by developers.
Change-Id: I7cf2dbeadecf994423d7e7591e85f18d2575cce8
BUG: 1522662
Signed-off-by: N Balachandran <nbalacha@redhat.com>
Summary:
- Per title, does a bounds check on the frame->op and bails from the
function if it's invalid preventing the crash
Test Plan: Prove tests
Reviewers: dph, jackl
Reviewed By: jackl
FB-commit-id: e67cc15
Change-Id: If1a5a9c0630573d4a6615050a9114ccf532551c7
BUG: 1522847
Signed-off-by: Kevin Vigor <kvigor@fb.com>
Reviewed-on: https://review.gluster.org/16847
CentOS-regression: Gluster Build System <jenkins@build.gluster.org>
NetBSD-regression: NetBSD Build System <jenkins@build.gluster.org>
Smoke: Gluster Build System <jenkins@build.gluster.org>
Reviewed-by: Shreyas Siravara <sshreyas@fb.com>
Summary:
- Reduce the amount of unnecessary timing calls
in iot_worker servicing.
- The current logic is unnecessarily accurate and
hurts performance for many small FOPS.
Change-Id: I6db4f1ad9a48d9d474bb251a2204969061021954
BUG: 1522950
Signed-off-by: Shreyas Siravara <sshreyas@fb.com>
Reviewed-on: http://review.gluster.org/16081
Smoke: Gluster Build System <jenkins@build.gluster.org>
NetBSD-regression: NetBSD Build System <jenkins@build.gluster.org>
CentOS-regression: Gluster Build System <jenkins@build.gluster.org>
Reviewed-by: Kevin Vigor <kvigor@fb.com>
specify ctx in gf_log_set_loglevel, instead of getting it from a thread
specific variable.
Change-Id: I498f826e8e32231235a6b0005026a27c327727fd
BUG: 1521213
Signed-off-by: Zhang Huan <zhanghuan@open-fs.com>
* Introduce xlator methods to allow dumping of metrics
* Separate options to get the metrics dumped in a path
Updates #168
Change-Id: I7df80df33b71d6f449f03c2332665b4a45f6ddf2
Signed-off-by: Amar Tumballi <amarts@redhat.com>
Problem: It had been a longtime request to implement put fop
in gluster. put fop in gluster may not have the exact sementics
of HTTP PUT, but can be easily extended to do so. The subsequent
patches, will contain more semantics on the put fop and its
guarentees.
Why compound fop framework is not used for put?
Compound fop framework currently doesn't allow compounding of
entry fop and inode fops, i.e. fops on multiple inodes cannot be
combined in compound fop.
Updates #353
Change-Id: Idb7891b3e056d46d570bb7e31bad1b6a28656ada
Signed-off-by: Poornima G <pgurusid@redhat.com>
To resolve a fd from client requests, need to take a mutex lock for the
fdtable to do the lookup. When a client is busy doing read and write,
the mutex lock could introduce contention. Therefore, use rwlock instead
of mutex to reduce the contention.
Change-Id: Ic833aed738a178a7ea1abafed7eb13814989d28c
BUG: 1518582
Signed-off-by: Zhang Huan <zhanghuan@open-fs.com>
each translator from now on can have just 1 symbol exported called
'xlator_api', which has all the required fields in it.
Updates: #164
Change-Id: I48d54f5ec59fee842b1d55877e3ac5e9ec9b6bdd
Signed-off-by: Amar Tumballi <amarts@redhat.com>
- call_stack_set_group() will take the ownership of passed
buffer from caller;
- to indicate the change, its signature is changed from
including the buffer directly to take a pointer to it;
- either the content of the buffer is copied to the
groups_small embedded buffer of the call stack, or
the buffer is set as groups_large member of the call
stack;
- the groups member of the call stack is set to,
respectively, groups_small or groups_large, according
to the memory management conventions of the call stack;
- the buffer address is overwritten with junk to effectively
prevent the caller from using it further on.
Also move call_stack_set_group to stack.c from stack.h
to prevent "defined but not used [-Wunused-function]"
warnings (not using it anymore in call_stack_alloc_group()
implementation, which saved us from this so far).
protocol/server: refactor gid_resolve()
In gid_resolve there are two cases:
either the gid_cache_lookup() call returns
a value or not. The result is caputured in
the agl variable, and throughout the function,
each particular stage of the implementation
comes with an agl and a no-agl variant.
In most cases this is explicitly indicated
via an
if (agl) {
...
} else {
...
}
but some of this branching are expressed via
goto constructs (obfuscating the fact we stated
above, that is, each particular stage having
an agl/no-agl variant).
In the current refactor, we bring the agl
conditional to the top, and present the
agl/non-agl implementations sequentially.
Also we take the opportunity to clean up and
fix the agl case:
- remove the spurious
gl.gl_list = agl->gl_list;
setting, as gl is not used in the agl caae
- populate the group list of call stack from
agl, fixing thus referred BUG.
Also fixes BUG: 1513920
Change-Id: I61f4574ba21969f7661b9ff0c9dce202b874025d
BUG: 1513928
Signed-off-by: Csaba Henk <csaba@redhat.com>
Problem:
FS sometimes doesn't give the expected return values. We need our common
functions to guard against this.
Example BUG: https://bugzilla.redhat.com/show_bug.cgi?id=864401
Fix:
When the return value is not as per specification, change the return value
to -1 and errno to EIO
BUG: 1469487
Change-Id: I14739ab2e5ae225b1a91438b87f8928af56f2934
Signed-off-by: Pranith Kumar K <pkarampu@redhat.com>
In this implementation, inodelk/entrylk will be tried for the subvols
given with trylock. In this attempt if all locks are obtained, then
inodelk is successful, otherwise, if it gets success on the first
available subvolume, then it will go for blocking lock, where as other
subvolumes will not try and this acts as tie-breaker.
Updates gluster/glusterfs#354
Change-Id: Ia2521b9ccb81a42bd6104ab21f610f761ba2b801
Signed-off-by: Pranith Kumar K <pkarampu@redhat.com>
As mentioned in bug1509189, there is a possible race
between gf_timer_cancel(), gf_timer_proc() and
gf_timer_registry_destroy() leading to use_after_free.
Problem:
1) gf_timer_proc() is called, locks reg, and gets an event.
It unlocks reg, and calls the callback.
2) Meanwhile gf_timer_registry_destroy() is called, and removes
reg from ctx, and joins on gf_timer_proc().
3) gf_timer_call_cancel() is called on the event being
processed. It cannot find reg (since it's been removed from reg),
so it frees event.
4) the callback returns into gf_timer_proc(), and it tries to free
event, but it's already free, so double free.
Solution:
The fix is to bail out in gf_timer_cancel() when registry
is not found. The logic behind this is that, gf_timer_cancel()
is called only on any existing event. That means there was a valid
registry earlier while creating that event. And the only reason
we cannot find that registry now is that it must have got set to
NULL when context cleanup is started.
Since gf_timer_proc() takes care of releasing all the remaining
events active on that registry, it seems safe to bail out
in gf_timer_cancel().
Change-Id: Ia9b088533141c3bb335eff2fe06b52d1575bb34f
BUG: 1509189
Reported-by: Daniel Gryniewicz <dang@redhat.com>
Signed-off-by: Soumya Koduri <skoduri@redhat.com>
Problem:Calling "gf_thread_create" without checking return value.
Fix:The return value is saved and checked if gf_thread_create fails.
Change-Id: Ibdaac1c90a1a8369e92ade50825598b041063da8
BUG: 789278
Signed-off-by: Subha sree Mohankumar <smohanku@redhat.com>
Problem:
Today the main users of client uuid are protocol layers, locks, leases.
Protocol layers requires each client uuid to be unique, even across
connects and disconnects. Locks and leases on the server side also use
the same client uid which changes across file migrations. Which makes the graph
switch and file migration tedious for locks and leases.
file migration across bricks becomes difficult as client uuid for the same
client, is different on the other brick.
The exact set of issues exists for leases as well.
Solution would be to introduce a constant in the client-uid string which
the locks and leases can use to identify the owner client across bricks.
Client uuid currently:
%s(ctx uuid)-%s(protocol client name)-%d(graph id)%s(setvolume count/reconnect count)
Proposed Client uuid:
"CTX_ID:%s-GRAPH_ID:%d-PID:%d-HOST:%s-PC_NAME:%s-RECON_NO:%s"
- CTX_ID: This is will be constant per client.
- GRAPH_ID, PID, HOST, PC_NAME(protocol client name), RECON_NO(setvolume count)
remains the same.
Change-Id: Ia81d57a9693207cd325d7b26aee4593fcbd6482c
BUG: 1369028
Signed-off-by: Susant Palai <spalai@redhat.com>
The 'setkey' will be used as the key by GD2 when setting the option
during volgen. 'setkey' also supports using varstrings.
This is mainly to be used for options, which use a different key for
'volume set' and in volfiles. For eg. the 'auth.*' options of
protocol/server.
The protocol/server xlator has been updated to make use of this for the
auth.allow and auth.reject options.
Updates #302
Change-Id: I1fd2fd69625c9db48595bd3f494c221625255169
Signed-off-by: Amar Tumballi <amarts@redhat.com>
This patch solves a detection problem in configure.ac that prevented
that compilation detects builtin __atomic or __sync functions.
It also adds more atomic types and support for other atomic functions.
An special case has been added to support 64-bit atomics on 32-bit
systems. The solution is to fallback to the mutex solution only for
64-bit atomics, but smaller atomic types will still take advantage
of builtins if available.
Change-Id: I6b9afc7cd6e66b28a33278715583552872278801
BUG: 1510397
Signed-off-by: Xavier Hernandez <jahernan@redhat.com>
Coverity Id:719,748,761
problem : The value of "ret" is overwritten in init_db and
add_conection_node
Change-Id: Iade8ca8d61c5e25e8c311b1375219f5f61d51bc3
BUG: 789278
Signed-off-by: Subha sree Mohankumar <smohanku@redhat.com>
Problem : Overwriting previous write to "ret" with value "EVENT_SEND_OK"
before itr can be used.
Fix: The value of ret is used in out.
Change-Id: I2cdb32e441c85c94de30de89a7a4121fd54d1acd
BUG: 789278
Signed-off-by: Subha sree Mohankumar <smohanku@redhat.com>
Make sure to handle these counters in STACK_WIND/UNWIND macro, and
keep the counters as part of xlator_t structure itself, to provide
infra to monitoring.
Updates #137
Change-Id: Ib54d45e2321c2b095dac5810c37e6cdffe1f71b7
Signed-off-by: Amar Tumballi <amarts@redhat.com>
This addresses two issues. First is that we currently keep
traversing the timer list even when it's no longer possible for
us to find a timer that's ready to fire. Second is that we sleep
a full second even when the next timer is due sooner.
Change-Id: I178a460d0176dbb45f972c62ee94c6df66d92ca5
Signed-off-by: Jeff Darcy <jdarcy@fb.com>
For achieving the above, needed below changes too.
* more sanity into how 'frame->op' is assigned.
* infra to have 'stats' as separate section in 'xlator_t' structure
Updates #137
Change-Id: I36679bf9577f3ed00a695b4e7d92870dcb3db8e1
Signed-off-by: Amar Tumballi <amarts@redhat.com>
A new option is added to allow independent configuration of eager
locking for regular files and non-regular files.
Change-Id: I8f80e46d36d8551011132b15c0fac549b7fb1c60
BUG: 1502610
Signed-off-by: Xavier Hernandez <jahernan@redhat.com>
This reduces the space used from four bytes to one, and allows
new code to use familiar C99 types/values interoperably with our
old cruft. It does *not* change current declarations or code;
that will be left for a separate - much larger - patch.
Updates: #80
Change-Id: I5baedd17d3fb05b38f0d8b8bb9dd62824475842e
Signed-off-by: Jeff Darcy <jdarcy@fb.com>
Summary:
- This diff changes all locations in the code to prefer inet6 family
instead of inet. This will allow change GlusterFS to operate
via IPv6 instead of IPv4 for all internal operations while still
being able to serve (FUSE or NFS) clients via IPv4.
- The changes apply to NFS as well.
- This diff ports D1892990, D1897341 & D1896522 to the 3.8 branch.
Test Plan: Prove tests!
Reviewers: dph, rwareing
Signed-off-by: Shreyas Siravara <sshreyas@fb.com>
Change-Id: I34fdaaeb33c194782255625e00616faf75d60c33
BUG: 1406898
Reviewed-on-3.8-fb: http://review.gluster.org/16059
Reviewed-by: Shreyas Siravara <sshreyas@fb.com>
Tested-by: Shreyas Siravara <sshreyas@fb.com>
If 'oldgroups' is null, it can lead to a crash.
Change-Id: I7a4f71256b07dc0ca3b988cad3694c21b150c99e
BUG: 789278
Signed-off-by: Amar Tumballi <amarts@redhat.com>
Problem:
In a multinode environment, if two of the op-sm transactions
are initiated on one of the receiver nodes at the same time,
there might be a possibility that glusterd may end up in
stale lock.
Solution:
During mgmt_v3_lock a registration is made to gf_timer_call_after
which release the lock after certain period of time
Change-Id: I16cc2e5186a2e8a5e35eca2468b031811e093843
BUG: 1499004
Signed-off-by: Gaurav Yadav <gyadav@redhat.com>
'make clean' does not cleanup everything, and some of the files get
cleaned too eagerly. Several files are being packaged in a 'make dist'
tarball, that get rebuild each time anyway.
Specifically, this change prevents
- libglusterfs/src/generator.pyc from laying around
- keeping rpc/xdr/gen/*.x symlinks
- modifying tests/basic/{fuse,gfapi}/Makefile each run
- including tests/env.rc and events/src/eventtypes.py in the tarball
Change-Id: I774dd1abf3a9d3b6a89b938cf6ee7d7792c59a82
BUG: 1501317
Reported-by: Patrick Matthäi <pmatthaei@debian.org>
Signed-off-by: Niels de Vos <ndevos@redhat.com>
Currently, server protocol's init and glusterd's option
validation methods are different, causing an issue. They
should be same for having consistent behavior
Updates #175
Change-Id: Ibbf9a18c7192b2d77f9b7675ae7da9b8d2fe5de4
Signed-off-by: Amar Tumballi <amarts@redhat.com>
Problem:
Commit ff075a3d6f9b142911d25c27fd209838782bfff0 disabled loading
client-io-threads for replicate volumes (it was set to on by default in
commit e068c1997314046658dd502e9118dab32decf879) due to performance
issues but in doing so, inadvertently failed to load the xlator even if
the user explicitly enabled the option using the volume set command.
This was despite returning returning sucess for the volume set.
Fix:
Modify the check in perfxl_option_handler() and add checks in volume
create/add-brick/remove-brick code paths, tying it all to
GD_OP_VERSION_3_12_2.
Change-Id: Ib612973a999a7da818cc926f5c2601b1f0794fcf
BUG: 1498570
Signed-off-by: Ravishankar N <ravishankar@redhat.com>
Existing EC code updates the xattr on the subvolume
in a sequential pattern resulting in very poor performance.
With this fix EC now updates the xattr on the subvolume
in parallel which improves the xattr update performance.
BUG: 1445663
Change-Id: I3fc40d66db0b88875ca96a9fa01002ba386c0486
Signed-off-by: Sunil Kumar Acharya <sheggodu@redhat.com>
Adding the implementation for the posix_do_futimes function which is
not complete in the current implementation and giving the ENOSYS error.
Change-Id: I9cfc95a7ea293b0a2df8efd4ac80d0120b3120e4
BUG: 1350406
Signed-off-by: karthik-us <ksubrahm@redhat.com>
Issue : "l <= 9223372036854775807L" is always true regardless of the values
of its operands. This occurs as the logical operand of "if".
Solution : Remove the comparison which always turns out to be true
Fix : The if-condition was removed and the body inside the same was retained.
Change-Id: Iba94d7f4f2dee85a180d10cdb7f7235b406cc400
BUG: 789278
Signed-off-by: Mohammed Azhar Padariyakam <mpadariy@redhat.com>
Issue: Event negative_return_fn: Function "dup(handle->fd)" returns a
negative number. Event negative_returns: "dup(handle->fd)" is passed
to a parameter that cannot be negative.
With this change value of dup(handle->fd) is stored in duped_fd & if
condition checks the value of duped_fd is non-negative.
Change-Id: I563d717108016d740ffa64fbe0929eb1e08c8f33
BUG: 789278
Signed-off-by: Girjesh Rajoria <grajoria@redhat.com>
Problem: Current code is not checking whether the pool pointer is null or not.
Solution: Updated the code to verify pool pointer.
Bug: 1496675
Change-Id: Ie1f2de4e4204fde15d2b1e3a966ea4c9e7b41534
Signed-off-by: Akarsha Rai <akrai@redhat.com>
The headerfile globals.h is recursively adding itself.
( globals.h -> xlator.h -> stack.h -> globals.h).
We are finding the source files which are including the header
file globals.h and removing the inclusion line.
I used git grep -l stack.h | xargs git grep globals.h --
to find out the files and removed the header file from all files
except libglusterfs/src/xlator.h and libglusterfs/src/Makefile.am
When I try to remove header file from libglusterfs/src/xlator.h
I'm getting some errors. In libglusterfs/src/Makefile.am it is
required for building RPMs.
Change-Id: I537218c09ade6d7ea51717768b26563a247daf60
BUG: 789278
Signed-off-by: Sanju Rakonde <srakonde@redhat.com>
Problem: xlator.h which is included at inode.h:32 includes itself.
Solution: Delete the include statement.
Change-Id: I304bbef5293a2fef0584773a1f8e61e838d7cbf1
BUG: 789278
Signed-off-by: Kamal Mohanan <kmohanan@redhat.com>
... and disable it by default.
This is because having it disabled seems to improve performance.
This could be due to the lock contention by the different epoll threads
on the circular buff lock in the fop cbks just before writing their response
to /dev/fuse.
Just to provide some data - wrt ovirt-gluster hyperconverged
environment, I saw an increase in IOPs by 12K with event-history
disabled for randrom read workload.
Usage:
mount -t glusterfs -o event-history=on $HOSTNAME:$VOLNAME $MOUNTPOINT
OR
glusterfs --event-history=on --volfile-server=$HOSTNAME --volfile-id=$VOLNAME $MOUNTPOINT
Change-Id: Ia533788d309c78688a315dc8cd04d30fad9e9485
BUG: 1467614
Signed-off-by: Krutika Dhananjay <kdhananj@redhat.com>
GlusterFS failed during make on openSUSE Tumbleweed with the following
error:
Making all in fdl
Making all in src
CC logdump.o
CC recon.o
CC fdl.lo
CC librecon.o
CC libfdl.o
CCLD gf_logdump
CCLD gf_recon
/usr/lib64/gcc/x86_64-suse-linux/7/../../../../lib64/libfl.so: undefined reference to `yylex'
collect2: error: ld returned 1 exit status
make[5]: *** [Makefile:618: gf_logdump] Error 1
make[5]: *** Waiting for unfinished jobs....
/usr/lib64/gcc/x86_64-suse-linux/7/../../../../lib64/libfl.so: undefined reference to `yylex'
collect2: error: ld returned 1 exit status
Reading through autoconf manual[1](see AC_PROG_LEX) reveals that LEXLIB
is automatically set to appropriate value for the system. The reference
to LEXLIB in automake file caused the above mentioned error on openSUSE.
In particular, we do not bother about LEXLIB hereafter.
[1] https://www.gnu.org/software/autoconf/manual/autoconf.html#Particular-Programs
Change-Id: I9bfce80c9654b2e3bfb393b08c25e8ad3d79e449
BUG: 1493133
Signed-off-by: Anoop C S <anoopcs@redhat.com>
run.c can be compiled into a standalone object file
or a demo program with the appropriate defines for
development purposes. This functionality was broken.
We fix it and also clean it up:
- call it "demo mode" not "test mode" as tests should
come with verification of the results of the invocation
which is not being done
- add comments to the source explaining the feature
- provide more comprehensive output in the demo program
This change does not affect standard compilation and
usage of run.c functionality.
BUG: 1454590
Change-Id: I1ea618e3262bf6a4d9f79f6b59209438d5163244
Signed-off-by: Csaba Henk <csaba@redhat.com>
Reviewed-on: https://review.gluster.org/17363
Smoke: Gluster Build System <jenkins@build.gluster.org>
Tested-by: Amar Tumballi <amarts@redhat.com>
NetBSD-regression: NetBSD Build System <jenkins@build.gluster.org>
CentOS-regression: Gluster Build System <jenkins@build.gluster.org>
Reviewed-by: Amar Tumballi <amarts@redhat.com>
The warning is specific to FreeBSD, since both Linux and NetBSD
do trigger a different codepath.
Change-Id: I3d2b374b6a39804942af076d7e0d130a76f869a2
BUG: 1488808
Signed-off-by: Michael Scherer <misc@redhat.com>
Reviewed-on: https://review.gluster.org/18214
Smoke: Gluster Build System <jenkins@build.gluster.org>
Reviewed-by: Niels de Vos <ndevos@redhat.com>
Tested-by: Niels de Vos <ndevos@redhat.com>
CentOS-regression: Gluster Build System <jenkins@build.gluster.org>
we register socket with EPOLLONESHOT, which means it has to be
explicitly added back through epoll_ctl to receive more
events. Normally we do this once the handler completes processing of
current event. But event_select_on_epoll is one asynchronous codepath
where socket can be added back for polling while an event on the same
socket is being processed. event_select_on_epoll has a check whether
an event is being processed in the form of slot->in_handler. But this
check is not sufficient enough to prevent parallel events as
slot->in_handler is not atomically incremented with respect to
reception of the event. This means following imaginary sequence of
events can happen:
* epoll_wait returns with a POLLERR - say POLLERR1 - on a socket
(sock1) associated with slot s1. socket_event_handle_pollerr is yet
to be invoked.
* an event_select_on called from __socket_ioq_churn which was called
in request/reply/msg submission codepath (as opposed to
__socket_ioq_churn called as part of POLLOUT handling - we cannot
receive a POLLOUT due to EPOLLONESHOT) adds back sock1 for polling.
* since sock1 was added back for polling in step 2 and our polling is
level-triggered, another thread picks up another POLLERR event - say
POLLERR2. socket_event_handler is invoked as part of processing
POLLERR2 and it completes execution setting priv->sock to -1.
* event_unregister_epoll called as part of __socket_reset due to
POLLERR1 would receive fd as -1 resulting in assert failure.
Also, since the first pollerr event has done rpc_transport_unref,
subsequent parallel events (not just pollerr, but other events too)
could be acting on a freed up transport too.
Change-Id: I5db755068e7890ec755b59f7a35a57da110339eb
BUG: 1486134
Signed-off-by: Raghavendra G <rgowdapp@redhat.com>
Reviewed-on: https://review.gluster.org/18129
Smoke: Gluster Build System <jenkins@build.gluster.org>
CentOS-regression: Gluster Build System <jenkins@build.gluster.org>
Reviewed-by: mohammed rafi kc <rkavunga@redhat.com>
Problem: Run-time crash is observed when attempting to memset() a zero
length buffer.
Solution: When GF_DISABLE_MEMPOOL is set, mem_get() gets translated to a
GF_MALLOC(). The size of the allocation does not need to relate to the
available (but uninitialized) global memory pools. It is fine to
allocate the exact amount of memory that was configured when the
mem-pool was created.
Change-Id: Iea0bff974bb771623a34d7a940e10cb0db0f90e1
BUG: 1481199
Reported-by: Milind Changire <mchangir@redhat.com>
Signed-off-by: Niels de Vos <ndevos@redhat.com>
Reviewed-on: https://review.gluster.org/18034
Smoke: Gluster Build System <jenkins@build.gluster.org>
CentOS-regression: Gluster Build System <jenkins@build.gluster.org>
Reviewed-by: Jeff Darcy <jeff@pl.atyp.us>
