Pádraig Brady <[email protected]> writes:

> On 15/04/2026 09:34, Lukáš Zaoral wrote:
>>> On 14. 4. 2026, at 23:07, Pádraig Brady <[email protected]> wrote:
>>>
>>> On 14/04/2026 13:09, Lukáš Zaoral wrote:
>>>> Do not compare only with the latest entry for given device id but also
>>>> all previously saved entries with the same id.
>>>> * src/df.c (struct devlist): Add next_same_dev struct member.
>>>> (filter_mount_list): Iterate over next_same_dev to find duplicates.
>>>> * tests/df/skip-duplicates.sh: Add test cases.
>>>
>>> This does look correct,
>>> though I'm a bit worried about quadratic behaviour
>>> in the case of many mounts sharing the same dev id.
>> Yes, I’m aware of the complexity but I couldn’t think of a better
>> way.
>> 
>>> Was there a distro bug to help base the practical considerations.
>> Yes, there is a distribution bug [1] but it is not urgent (the link
>> should
>> be public).  This problem was found during the investigation of other
>> issues by my colleagues from support.
>> This ticket sat on the bottom of my backlog for a very long time and
>> I
>> have no plans to fix it in RHEL at this point but since I’ve already
>> made a patch, I wanted to send it to upstream.
>> Regards,
>> Lukas
>> [1]: https://redhat.atlassian.net/browse/RHEL-5649
>
> OK I included that reference and pushed.
> Let's not over engineer it for the uncommon case.
> We can revisit if it becomes an issue.

FWIW, my Gnome Terminal and Emacs sessions start to sporadically become
unresponsive after creating almost 5000 bind mounts. All of the 'df'
commands I executed finished reasonably quickly, though.

Collin

Reply via email to