2022-10-31 15:59:02

by Jeff Layton

[permalink] [raw]
Subject: [PATCH v2] nfsd: fix net-namespace logic in __nfsd_file_cache_purge

If the namespace doesn't match the one in "net", then we'll continue,
but that doesn't cause another rhashtable_walk_next call, so it will
loop infinitely.

Fixes: ce502f81ba88 ("NFSD: Convert the filecache to use rhashtable")
Reported-by: Petr Vorel <[email protected]>
Signed-off-by: Jeff Layton <[email protected]>
---
fs/nfsd/filecache.c | 5 ++---
1 file changed, 2 insertions(+), 3 deletions(-)

The v1 patch applies cleanly to v6.0, but not to Chuck's for-next
branch. This one should be suitable there.
diff --git a/fs/nfsd/filecache.c b/fs/nfsd/filecache.c
index 98c6b5f51bc8..4a8aa7cd8354 100644
--- a/fs/nfsd/filecache.c
+++ b/fs/nfsd/filecache.c
@@ -890,9 +890,8 @@ __nfsd_file_cache_purge(struct net *net)

nf = rhashtable_walk_next(&iter);
while (!IS_ERR_OR_NULL(nf)) {
- if (net && nf->nf_net != net)
- continue;
- nfsd_file_unhash_and_dispose(nf, &dispose);
+ if (!net || nf->nf_net == net)
+ nfsd_file_unhash_and_dispose(nf, &dispose);
nf = rhashtable_walk_next(&iter);
}

--
2.38.1



2022-10-31 17:15:21

by Chuck Lever III

[permalink] [raw]
Subject: Re: [PATCH v2] nfsd: fix net-namespace logic in __nfsd_file_cache_purge



> On Oct 31, 2022, at 11:49 AM, Jeff Layton <[email protected]> wrote:
>
> If the namespace doesn't match the one in "net", then we'll continue,
> but that doesn't cause another rhashtable_walk_next call, so it will
> loop infinitely.
>
> Fixes: ce502f81ba88 ("NFSD: Convert the filecache to use rhashtable")
> Reported-by: Petr Vorel <[email protected]>
> Signed-off-by: Jeff Layton <[email protected]>
> ---
> fs/nfsd/filecache.c | 5 ++---
> 1 file changed, 2 insertions(+), 3 deletions(-)

Thank you! Applied and pushed to for-rc. I'll send a PR in a few days.


> The v1 patch applies cleanly to v6.0, but not to Chuck's for-next
> branch. This one should be suitable there.
> diff --git a/fs/nfsd/filecache.c b/fs/nfsd/filecache.c
> index 98c6b5f51bc8..4a8aa7cd8354 100644
> --- a/fs/nfsd/filecache.c
> +++ b/fs/nfsd/filecache.c
> @@ -890,9 +890,8 @@ __nfsd_file_cache_purge(struct net *net)
>
> nf = rhashtable_walk_next(&iter);
> while (!IS_ERR_OR_NULL(nf)) {
> - if (net && nf->nf_net != net)
> - continue;
> - nfsd_file_unhash_and_dispose(nf, &dispose);
> + if (!net || nf->nf_net == net)
> + nfsd_file_unhash_and_dispose(nf, &dispose);
> nf = rhashtable_walk_next(&iter);
> }
>
> --
> 2.38.1
>

--
Chuck Lever




2022-10-31 18:45:40

by Petr Vorel

[permalink] [raw]
Subject: Re: [PATCH v2] nfsd: fix net-namespace logic in __nfsd_file_cache_purge



> > On Oct 31, 2022, at 11:49 AM, Jeff Layton <[email protected]> wrote:

> > If the namespace doesn't match the one in "net", then we'll continue,
> > but that doesn't cause another rhashtable_walk_next call, so it will
> > loop infinitely.

> > Fixes: ce502f81ba88 ("NFSD: Convert the filecache to use rhashtable")
> > Reported-by: Petr Vorel <[email protected]>
> > Signed-off-by: Jeff Layton <[email protected]>
> > ---
> > fs/nfsd/filecache.c | 5 ++---
> > 1 file changed, 2 insertions(+), 3 deletions(-)

> Thank you! Applied and pushed to for-rc. I'll send a PR in a few days.

Thanks for a quick action!

What a shame you didn't put link to the report, which contains reproducer.
https://lore.kernel.org/ltp/Y1%2FP8gDAcWC%2F+VR3@pevik/

Kind regards,
Petr

> > The v1 patch applies cleanly to v6.0, but not to Chuck's for-next
> > branch. This one should be suitable there.
> > diff --git a/fs/nfsd/filecache.c b/fs/nfsd/filecache.c
> > index 98c6b5f51bc8..4a8aa7cd8354 100644
> > --- a/fs/nfsd/filecache.c
> > +++ b/fs/nfsd/filecache.c
> > @@ -890,9 +890,8 @@ __nfsd_file_cache_purge(struct net *net)

> > nf = rhashtable_walk_next(&iter);
> > while (!IS_ERR_OR_NULL(nf)) {
> > - if (net && nf->nf_net != net)
> > - continue;
> > - nfsd_file_unhash_and_dispose(nf, &dispose);
> > + if (!net || nf->nf_net == net)
> > + nfsd_file_unhash_and_dispose(nf, &dispose);
> > nf = rhashtable_walk_next(&iter);
> > }

> > --
> > 2.38.1

2022-10-31 18:50:05

by Jeff Layton

[permalink] [raw]
Subject: Re: [PATCH v2] nfsd: fix net-namespace logic in __nfsd_file_cache_purge

On Mon, 2022-10-31 at 19:45 +0100, Petr Vorel wrote:
>
> > > On Oct 31, 2022, at 11:49 AM, Jeff Layton <[email protected]> wrote:
>
> > > If the namespace doesn't match the one in "net", then we'll continue,
> > > but that doesn't cause another rhashtable_walk_next call, so it will
> > > loop infinitely.
>
> > > Fixes: ce502f81ba88 ("NFSD: Convert the filecache to use rhashtable")
> > > Reported-by: Petr Vorel <[email protected]>
> > > Signed-off-by: Jeff Layton <[email protected]>
> > > ---
> > > fs/nfsd/filecache.c | 5 ++---
> > > 1 file changed, 2 insertions(+), 3 deletions(-)
>
> > Thank you! Applied and pushed to for-rc. I'll send a PR in a few days.
>
> Thanks for a quick action!
>

No problem.

> What a shame you didn't put link to the report, which contains reproducer.
> https://lore.kernel.org/ltp/Y1%2FP8gDAcWC%2F+VR3@pevik/
>
> Kind regards,
> Petr
>

Chuck, could you add that link to the changelog when you merge it?

Thanks,
Jeff

> > > The v1 patch applies cleanly to v6.0, but not to Chuck's for-next
> > > branch. This one should be suitable there.
> > > diff --git a/fs/nfsd/filecache.c b/fs/nfsd/filecache.c
> > > index 98c6b5f51bc8..4a8aa7cd8354 100644
> > > --- a/fs/nfsd/filecache.c
> > > +++ b/fs/nfsd/filecache.c
> > > @@ -890,9 +890,8 @@ __nfsd_file_cache_purge(struct net *net)
>
> > > nf = rhashtable_walk_next(&iter);
> > > while (!IS_ERR_OR_NULL(nf)) {
> > > - if (net && nf->nf_net != net)
> > > - continue;
> > > - nfsd_file_unhash_and_dispose(nf, &dispose);
> > > + if (!net || nf->nf_net == net)
> > > + nfsd_file_unhash_and_dispose(nf, &dispose);
> > > nf = rhashtable_walk_next(&iter);
> > > }
>
> > > --
> > > 2.38.1

--
Jeff Layton <[email protected]>

2022-10-31 18:51:55

by Chuck Lever III

[permalink] [raw]
Subject: Re: [PATCH v2] nfsd: fix net-namespace logic in __nfsd_file_cache_purge



> On Oct 31, 2022, at 2:46 PM, Jeff Layton <[email protected]> wrote:
>
> On Mon, 2022-10-31 at 19:45 +0100, Petr Vorel wrote:
>>
>>>> On Oct 31, 2022, at 11:49 AM, Jeff Layton <[email protected]> wrote:
>>
>>>> If the namespace doesn't match the one in "net", then we'll continue,
>>>> but that doesn't cause another rhashtable_walk_next call, so it will
>>>> loop infinitely.
>>
>>>> Fixes: ce502f81ba88 ("NFSD: Convert the filecache to use rhashtable")
>>>> Reported-by: Petr Vorel <[email protected]>
>>>> Signed-off-by: Jeff Layton <[email protected]>
>>>> ---
>>>> fs/nfsd/filecache.c | 5 ++---
>>>> 1 file changed, 2 insertions(+), 3 deletions(-)
>>
>>> Thank you! Applied and pushed to for-rc. I'll send a PR in a few days.
>>
>> Thanks for a quick action!
>>
>
> No problem.
>
>> What a shame you didn't put link to the report, which contains reproducer.
>> https://lore.kernel.org/ltp/Y1%2FP8gDAcWC%2F+VR3@pevik/
>>
>> Kind regards,
>> Petr
>>
>
> Chuck, could you add that link to the changelog when you merge it?

In progress, I'll push the update in a moment.


>
> Thanks,
> Jeff
>
>>>> The v1 patch applies cleanly to v6.0, but not to Chuck's for-next
>>>> branch. This one should be suitable there.
>>>> diff --git a/fs/nfsd/filecache.c b/fs/nfsd/filecache.c
>>>> index 98c6b5f51bc8..4a8aa7cd8354 100644
>>>> --- a/fs/nfsd/filecache.c
>>>> +++ b/fs/nfsd/filecache.c
>>>> @@ -890,9 +890,8 @@ __nfsd_file_cache_purge(struct net *net)
>>
>>>> nf = rhashtable_walk_next(&iter);
>>>> while (!IS_ERR_OR_NULL(nf)) {
>>>> - if (net && nf->nf_net != net)
>>>> - continue;
>>>> - nfsd_file_unhash_and_dispose(nf, &dispose);
>>>> + if (!net || nf->nf_net == net)
>>>> + nfsd_file_unhash_and_dispose(nf, &dispose);
>>>> nf = rhashtable_walk_next(&iter);
>>>> }
>>
>>>> --
>>>> 2.38.1
>
> --
> Jeff Layton <[email protected]>

--
Chuck Lever




2022-10-31 19:14:02

by Petr Vorel

[permalink] [raw]
Subject: Re: [PATCH v2] nfsd: fix net-namespace logic in __nfsd_file_cache_purge



> > On Oct 31, 2022, at 2:46 PM, Jeff Layton <[email protected]> wrote:

> > On Mon, 2022-10-31 at 19:45 +0100, Petr Vorel wrote:

> >>>> On Oct 31, 2022, at 11:49 AM, Jeff Layton <[email protected]> wrote:

> >>>> If the namespace doesn't match the one in "net", then we'll continue,
> >>>> but that doesn't cause another rhashtable_walk_next call, so it will
> >>>> loop infinitely.

> >>>> Fixes: ce502f81ba88 ("NFSD: Convert the filecache to use rhashtable")
> >>>> Reported-by: Petr Vorel <[email protected]>
> >>>> Signed-off-by: Jeff Layton <[email protected]>
> >>>> ---
> >>>> fs/nfsd/filecache.c | 5 ++---
> >>>> 1 file changed, 2 insertions(+), 3 deletions(-)

> >>> Thank you! Applied and pushed to for-rc. I'll send a PR in a few days.

> >> Thanks for a quick action!


> > No problem.

> >> What a shame you didn't put link to the report, which contains reproducer.
> >> https://lore.kernel.org/ltp/Y1%2FP8gDAcWC%2F+VR3@pevik/

> >> Kind regards,
> >> Petr


> > Chuck, could you add that link to the changelog when you merge it?

> In progress, I'll push the update in a moment.

Thanks a lot!

Kind regards,
Petr



> > Thanks,
> > Jeff

> >>>> The v1 patch applies cleanly to v6.0, but not to Chuck's for-next
> >>>> branch. This one should be suitable there.
> >>>> diff --git a/fs/nfsd/filecache.c b/fs/nfsd/filecache.c
> >>>> index 98c6b5f51bc8..4a8aa7cd8354 100644
> >>>> --- a/fs/nfsd/filecache.c
> >>>> +++ b/fs/nfsd/filecache.c
> >>>> @@ -890,9 +890,8 @@ __nfsd_file_cache_purge(struct net *net)

> >>>> nf = rhashtable_walk_next(&iter);
> >>>> while (!IS_ERR_OR_NULL(nf)) {
> >>>> - if (net && nf->nf_net != net)
> >>>> - continue;
> >>>> - nfsd_file_unhash_and_dispose(nf, &dispose);
> >>>> + if (!net || nf->nf_net == net)
> >>>> + nfsd_file_unhash_and_dispose(nf, &dispose);
> >>>> nf = rhashtable_walk_next(&iter);
> >>>> }

> >>>> --
> >>>> 2.38.1

> > --
> > Jeff Layton <[email protected]>

2022-11-04 21:26:40

by Petr Vorel

[permalink] [raw]
Subject: Re: [PATCH v2] nfsd: fix net-namespace logic in __nfsd_file_cache_purge

Hi all,

> If the namespace doesn't match the one in "net", then we'll continue,
> but that doesn't cause another rhashtable_walk_next call, so it will
> loop infinitely.

> Fixes: ce502f81ba88 ("NFSD: Convert the filecache to use rhashtable")

Adding this to regression tracker.
https://linux-regtracking.leemhuis.info/tracked-regression/

#regzbot ^introduced ce502f81ba88
#regzbot ignore-activity

Kind regards,
Petr

> Reported-by: Petr Vorel <[email protected]>
> Signed-off-by: Jeff Layton <[email protected]>
> ---
> fs/nfsd/filecache.c | 5 ++---
> 1 file changed, 2 insertions(+), 3 deletions(-)

> The v1 patch applies cleanly to v6.0, but not to Chuck's for-next
> branch. This one should be suitable there.
> diff --git a/fs/nfsd/filecache.c b/fs/nfsd/filecache.c
> index 98c6b5f51bc8..4a8aa7cd8354 100644
> --- a/fs/nfsd/filecache.c
> +++ b/fs/nfsd/filecache.c
> @@ -890,9 +890,8 @@ __nfsd_file_cache_purge(struct net *net)

> nf = rhashtable_walk_next(&iter);
> while (!IS_ERR_OR_NULL(nf)) {
> - if (net && nf->nf_net != net)
> - continue;
> - nfsd_file_unhash_and_dispose(nf, &dispose);
> + if (!net || nf->nf_net == net)
> + nfsd_file_unhash_and_dispose(nf, &dispose);
> nf = rhashtable_walk_next(&iter);
> }

2022-11-09 16:04:14

by Thorsten Leemhuis

[permalink] [raw]
Subject: Re: [PATCH v2] nfsd: fix net-namespace logic in __nfsd_file_cache_purge #forregzbot

[Note: this mail is primarily send for documentation purposes and/or for
regzbot, my Linux kernel regression tracking bot. That's why I removed
most or all folks from the list of recipients, but left any that looked
like a mailing lists. These mails usually contain '#forregzbot' in the
subject, to make them easy to spot and filter out.]

On 04.11.22 22:21, Petr Vorel wrote:
> Hi all,
>
>> If the namespace doesn't match the one in "net", then we'll continue,
>> but that doesn't cause another rhashtable_walk_next call, so it will
>> loop infinitely.
>
>> Fixes: ce502f81ba88 ("NFSD: Convert the filecache to use rhashtable")
>
> Adding this to regression tracker.
> https://linux-regtracking.leemhuis.info/tracked-regression/
>
> #regzbot ^introduced ce502f81ba88
> #regzbot ignore-activity

#regzbot fixed-by: d3aefd2b29ff5