[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
bug#21460: Race condition in tests/tail-2/assert.sh
From: |
Pádraig Brady |
Subject: |
bug#21460: Race condition in tests/tail-2/assert.sh |
Date: |
Fri, 2 Oct 2015 16:50:22 +0100 |
User-agent: |
Mozilla/5.0 (X11; Linux x86_64; rv:38.0) Gecko/20100101 Thunderbird/38.2.0 |
On 12/09/15 03:22, Pádraig Brady wrote:
> On 12/09/15 02:09, Paul Eggert wrote:
>> Pádraig Brady wrote:
>>> + else if (new_stats.st_nlink == 0) /* XXX: what about multi-linked files.
>>> */
>>
>> That comment was my thought exactly. It appears to be a kernel bug that
>> really
>> needs to get fixed in the kernel; there just doesn't seem to be a reliable
>> workaround in user space.
>
> I double checked with kernel guys, and Al Viro
> essentially said the inode and directory operations are independent.
> https://lkml.org/lkml/2015/9/11/790
>
> So we probably need to handle the IN_DELETE event
> on the directory to cater for this race, as done in:
> https://lists.gnu.org/archive/html/coreutils/2015-07/msg00015.html
I'll apply the attached later.
thanks,
Pádraig
tail-unlink-notification-race.patch
Description: Text Data
[Prev in Thread] |
Current Thread |
[Next in Thread] |
- bug#21460: Race condition in tests/tail-2/assert.sh,
Pádraig Brady <=