On 08/03/2016 06:04 PM, Grygorii Strashko wrote:

Hi Vladimir,

On 08/03/2016 03:06 PM, Vladimir Murzin wrote:

quoted

On 03/08/16 12:41, Grygorii Strashko wrote:

quoted

We observe Kernel boot failure while running NFS boot stress test (1000 iterations):
- Linux version 4.7.0 

I'd like to pay your attention that this issue also reproducible with
Kernel 4.7.0!
The same can be seen from the log I've provided in first e-mail:
[    0.000000] Linux version 4.7.0 (lcpdbld at dflsdit-build06.dal.design.ti.com) (gcc version 4.9.3 20150413 (prerelease) (Linaro GCC 4.9-2015.05) ) #1 SMP Fri Jul 29 17:41:27 CDT 2016


I've not run the test with current master at it's not been tagged yet.

Still in progress. rc1 unstable on my platforms due to other issues :(

quoted

quoted

- am335x-evm (TI AM335x EVM)
- failure rate 10-20 times per test.
Originally this issue was reproduced using TI Kernel 4.4
( git://git.ti.com/ti-linux-kernel/ti-linux-kernel.git, branch: ti-linux-4.4.y)
on both am335x-evm and am57xx-beagle-x15(am57xx-evm) platforms.
This issues has not been reproduced with TI Kernel 4.1 before.

The SysRq shows that system stuck in nfs_fs_mount()

[  207.904632] [<c07ab34c>] (schedule) from [<c0783554>] (rpc_wait_bit_killable+0x2c/0xd8)
[  207.912996] [<c0783554>] (rpc_wait_bit_killable) from [<c07ab7f0>] (__wait_on_bit+0x84/0xc0)
[  207.921812] [<c07ab7f0>] (__wait_on_bit) from [<c07ab890>] (out_of_line_wait_on_bit+0x64/0x70)
[  207.930810] [<c07ab890>] (out_of_line_wait_on_bit) from [<c07843f4>] (__rpc_execute+0x18c/0x544)
[  207.939988] [<c07843f4>] (__rpc_execute) from [<c0779f24>] (rpc_run_task+0x13c/0x158)
[  207.948166] [<c0779f24>] (rpc_run_task) from [<c0779f84>] (rpc_call_sync+0x44/0xc4)
[  207.956163] [<c0779f84>] (rpc_call_sync) from [<c077a04c>] (rpc_ping+0x48/0x68)
[  207.963796] [<c077a04c>] (rpc_ping) from [<c077a158>] (rpc_create_xprt+0xec/0x164)
[  207.971702] [<c077a158>] (rpc_create_xprt) from [<c077a2c0>] (rpc_create+0xf0/0x1a0)
[  207.979794] [<c077a2c0>] (rpc_create) from [<c0393088>] (nfs_create_rpc_client+0xd4/0xec)
[  207.988338] [<c0393088>] (nfs_create_rpc_client) from [<c0394d10>] (nfs_init_client+0x20/0x78)
[  207.997332] [<c0394d10>] (nfs_init_client) from [<c03949d4>] (nfs_create_server+0xa0/0x3bc)
[  208.006057] [<c03949d4>] (nfs_create_server) from [<c03b197c>] (nfs3_create_server+0x8/0x20)
[  208.014879] [<c03b197c>] (nfs3_create_server) from [<c03a34c4>] (nfs_try_mount+0xc4/0x1f0)
[  208.023513] [<c03a34c4>] (nfs_try_mount) from [<c03a2c48>] (nfs_fs_mount+0x290/0x910)
[  208.031702] [<c03a2c48>] (nfs_fs_mount) from [<c0294d24>] (mount_fs+0x44/0x168)

Has anyone else seen this issue?

I'd be appreciated for any help or advice related to this issue?

I did not look at details, but because it is 4.4 and __wait_on_bit
showed up you might want to look at [1]

[1] https://lkml.org/lkml/2015/11/20/472

Thanks. I'll take a look.

I've checked this thread and all three commits mentioned there are present in K4.4

=3.17

commit 743162013d40  sched: Remove proliferation of wait_on_bit() action functions

=4.4

commit 68985633bccb  sched/wait: Fix signal handling in bit wait helpers

=4.4

commit dfd01f026058  sched/wait: Fix the signal handling fix


Also, It seems first patch, probably, has copy-past error.
I'm not sure and it may be that patch is correct :)
Any way, It doesn't help with this issue if I use wait_on_bit_lock_io in nfs_page_group_lock().

743162013d40 ("sched: Remove proliferation of wait_on_bit() action functions")
-- does:

diff --git a/fs/nfs/pagelist.c b/fs/nfs/pagelist.c
index b6ee3a6..6104d35 100644
--- a/fs/nfs/pagelist.c
+++ b/fs/nfs/pagelist.c

@@ -138,12 +138,6 @@ nfs_iocounter_wait(struct nfs_io_counter *c)
        return __nfs_iocounter_wait(c);
 }
 
-static int nfs_wait_bit_uninterruptible(void *word)
-{
-       io_schedule();
-       return 0;
-}
-
 /*
  * nfs_page_group_lock - lock the head of the page group
  * @req - request in group that is to be locked

@@ -158,7 +152,6 @@ nfs_page_group_lock(struct nfs_page *req)
        WARN_ON_ONCE(head != head->wb_head);
 
        wait_on_bit_lock(&head->wb_flags, PG_HEADLOCK,
-                       nfs_wait_bit_uninterruptible,

[GS] But it seems should be wait_on_bit_lock_io() <----

                        TASK_UNINTERRUPTIBLE);
 }
 

@@ -425,9 +418,8 @@ void nfs_release_request(struct nfs_page *req)
 int
 nfs_wait_on_request(struct nfs_page *req)
 {
-       return wait_on_bit(&req->wb_flags, PG_BUSY,
-                       nfs_wait_bit_uninterruptible,
-                       TASK_UNINTERRUPTIBLE);
+       return wait_on_bit_io(&req->wb_flags, PG_BUSY,
+                             TASK_UNINTERRUPTIBLE);
 }

-- 
regards,
-grygorii