Re: v4.4-rc3 ext4 memory leak

[Bart Van Assche <bart.vanassche@sandisk.com>]

On 12/04/2015 07:22 AM, Theodore Ts'o wrote:
> On Thu, Dec 03, 2015 at 02:21:31PM -0800, Bart Van Assche wrote:
>> Hello,
>>
>> Running xfstests on top of an ext4 filesystem causes kmemleak to report the
>> following memory leak multiple times:
>>
>> unreferenced object 0xffff880408709730 (size 96):
>>    comm "fsstress", pid 35588, jiffies 4295006079 (age 11865.680s)
>>    hex dump (first 32 bytes):
>>      00 00 00 00 00 00 00 00 00 00 01 00 00 00 00 00  ................
>>      00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
>>    backtrace:
>>      [<ffffffff8151056a>] kmemleak_alloc+0x7a/0xc0
>>      [<ffffffff81165e6e>] __kmalloc+0xde/0x140
>>      [<ffffffffa022472e>] ext4_find_extent+0x26e/0x310 [ext4]
>>      [<ffffffffa0227f15>] ext4_insert_range+0x255/0x3c0 [ext4]
>>      [<ffffffffa022aca1>] ext4_fallocate+0x1f1/0x210 [ext4]
>>      [<ffffffff81175fb6>] vfs_fallocate+0x156/0x1e0
>>      [<ffffffff8117608e>] SyS_fallocate+0x4e/0x80
>>      [<ffffffff8151b617>] entry_SYSCALL_64_fastpath+0x12/0x6f
>>      [<ffffffffffffffff>] 0xffffffffffffffff
>
> Is this something new?   i..e, was this reproducible on v4.3?
>
> And can you isolate this to a specific test in xfstests?

Hello Ted,

So far I only ran xfstests against the 4.4-rc3 kernel so I do not know 
whether this also occurred with previous kernel versions.

kmemleak scans periodically for memory leaks. kmemleak started reporting 
these leaks after test generic/037 had started. So I think one of the 
first 37 tests triggered this leak.

Bart.