[PATCH] nfsd: destroy percpu stats counters after reply cache #5.11.0-rc5

[PATCH] nfsd: destroy percpu stats counters after reply cache #5.11.0-rc5

[Julian Schroeder]

From: Julian Schroeder <jumaco@amazon.com>
Date: Fri, 20 May 2022 18:33:27 +0000
Subject: [PATCH] nfsd: destroy percpu stats counters after reply cache
 shutdown
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Upon nfsd shutdown any pending DRC cache is freed. DRC cache use is
tracked via a percpu counter. In the current code the percpu counter
is destroyed before. If any pending cache is still present,
percpu_counter_add is called with a percpu counter==NULL. This causes
a kernel crash.
The solution is to destroy the percpu counter after the cache is freed.
Fixes: e567b98ce9a4b (“nfsd: protect concurrent access to nfsd stats counters”)
Signed-off-by: Julian Schroeder <jumaco@amazon.com>
---
 fs/nfsd/nfscache.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/fs/nfsd/nfscache.c b/fs/nfsd/nfscache.c
index 0b3f12aa37ff..7da88bdc0d6c 100644
--- a/fs/nfsd/nfscache.c
+++ b/fs/nfsd/nfscache.c
@@ -206,7 +206,6 @@ void nfsd_reply_cache_shutdown(struct nfsd_net *nn)
        struct svc_cacherep     *rp;
        unsigned int i;
 
-       nfsd_reply_cache_stats_destroy(nn);
        unregister_shrinker(&nn->nfsd_reply_cache_shrinker);
 
        for (i = 0; i < nn->drc_hashsize; i++) {
@@ -217,6 +216,7 @@ void nfsd_reply_cache_shutdown(struct nfsd_net *nn)
                                                                        rp, nn);
                }
        }
+       nfsd_reply_cache_stats_destroy(nn);
 
        kvfree(nn->drc_hashtbl);
        nn->drc_hashtbl = NULL;
-- 
2.32.0

Re: [PATCH] nfsd: destroy percpu stats counters after reply cache #5.11.0-rc5

[Greg KH]

On Mon, May 23, 2022 at 09:11:52PM +0000, Julian Schroeder wrote:
> From: Julian Schroeder <jumaco@amazon.com>
> Date: Fri, 20 May 2022 18:33:27 +0000
> Subject: [PATCH] nfsd: destroy percpu stats counters after reply cache
>  shutdown
> MIME-Version: 1.0
> Content-Type: text/plain; charset=UTF-8
> Content-Transfer-Encoding: 8bit
> Upon nfsd shutdown any pending DRC cache is freed. DRC cache use is
> tracked via a percpu counter. In the current code the percpu counter
> is destroyed before. If any pending cache is still present,
> percpu_counter_add is called with a percpu counter==NULL. This causes
> a kernel crash.
> The solution is to destroy the percpu counter after the cache is freed.
> Fixes: e567b98ce9a4b (“nfsd: protect concurrent access to nfsd stats counters”)
> Signed-off-by: Julian Schroeder <jumaco@amazon.com>
> ---
>  fs/nfsd/nfscache.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> diff --git a/fs/nfsd/nfscache.c b/fs/nfsd/nfscache.c
> index 0b3f12aa37ff..7da88bdc0d6c 100644
> --- a/fs/nfsd/nfscache.c
> +++ b/fs/nfsd/nfscache.c
> @@ -206,7 +206,6 @@ void nfsd_reply_cache_shutdown(struct nfsd_net *nn)
>         struct svc_cacherep     *rp;
>         unsigned int i;
>  
> -       nfsd_reply_cache_stats_destroy(nn);
>         unregister_shrinker(&nn->nfsd_reply_cache_shrinker);
>  
>         for (i = 0; i < nn->drc_hashsize; i++) {
> @@ -217,6 +216,7 @@ void nfsd_reply_cache_shutdown(struct nfsd_net *nn)
>                                                                         rp, nn);
>                 }
>         }
> +       nfsd_reply_cache_stats_destroy(nn);
>  
>         kvfree(nn->drc_hashtbl);
>         nn->drc_hashtbl = NULL;
> -- 
> 2.32.0
> 

What is the git commit id of this in Linus's tree?

And this patch is totally damaged with whitespace and can not be applied
at all :(

Please fix it up and submit it again.

thanks,

greg k-h

Re: [PATCH] nfsd: destroy percpu stats counters after reply cache #5.11.0-rc5

[Schroeder, Julian]

The patch made it into linux next. commit fd5e363eac77e.
I sent the patch to stable again via git send-email.
Thanks,
julian

Re: [PATCH] nfsd: destroy percpu stats counters after reply cache #5.11.0-rc5

[Greg KH]

On Fri, May 27, 2022 at 07:39:36PM +0000, Schroeder, Julian wrote:
> The patch made it into linux next. commit fd5e363eac77e.
> I sent the patch to stable again via git send-email.

I have no context here at all, sorry.

What are we supposed to do?

confused,

greg k-h

Re: [PATCH] nfsd: destroy percpu stats counters after reply cache #5.11.0-rc5

[Schroeder, Julian]

>> From: Julian Schroeder <jumaco@amazon.com>
>> Date: Fri, 20 May 2022 18:33:27 +0000
>> Subject: [PATCH] nfsd: destroy percpu stats counters after reply cache
>>  shutdown
>> MIME-Version: 1.0
>> Content-Type: text/plain; charset=UTF-8
>> Content-Transfer-Encoding: 8bit
>> Upon nfsd shutdown any pending DRC cache is freed. DRC cache use is
>> tracked via a percpu counter. In the current code the percpu counter
>> is destroyed before. If any pending cache is still present,
>> percpu_counter_add is called with a percpu counter==NULL. This causes
>> a kernel crash.
>> The solution is to destroy the percpu counter after the cache is freed.
>> Fixes: e567b98ce9a4b (âEURoenfsd: protect concurrent access to nfsd stats countersâEUR)
>> Signed-off-by: Julian Schroeder <jumaco@amazon.com>
>> ---
>>  fs/nfsd/nfscache.c | 2 +-
>>  1 file changed, 1 insertion(+), 1 deletion(-)
>> diff --git a/fs/nfsd/nfscache.c b/fs/nfsd/nfscache.c
>> index 0b3f12aa37ff..7da88bdc0d6c 100644
>> --- a/fs/nfsd/nfscache.c
>> +++ b/fs/nfsd/nfscache.c
>> @@ -206,7 +206,6 @@ void nfsd_reply_cache_shutdown(struct nfsd_net *nn)
>>         struct svc_cacherep     *rp;
>>         unsigned int i;
>>
>> -       nfsd_reply_cache_stats_destroy(nn);
>>         unregister_shrinker(&nn->nfsd_reply_cache_shrinker);
>>
>>         for (i = 0; i < nn->drc_hashsize; i++) {
>> @@ -217,6 +216,7 @@ void nfsd_reply_cache_shutdown(struct nfsd_net *nn)
>>                                                                         rp, nn);
>>                 }
>>         }
>> +       nfsd_reply_cache_stats_destroy(nn);
>>
>>         kvfree(nn->drc_hashtbl);
>>         nn->drc_hashtbl = NULL;
>> --
>> 2.32.0
>>

>What is the git commit id of this in Linus's tree?

>And this patch is totally damaged with whitespace and can not be applied
>at all :(

>Please fix it up and submit it again.

>thanks,

>greg k-h

The patch made it into linux next. commit fd5e363eac77e.
I sent the patch to stable again via git send-email (https://www.spinics.net/lists/stable/msg560818.html)

Thanks,
Julian.

Re: [PATCH] nfsd: destroy percpu stats counters after reply cache #5.11.0-rc5

[Greg KH]

On Sat, May 28, 2022 at 01:41:51PM +0000, Schroeder, Julian wrote:
> >> From: Julian Schroeder <jumaco@amazon.com>
> >> Date: Fri, 20 May 2022 18:33:27 +0000
> >> Subject: [PATCH] nfsd: destroy percpu stats counters after reply cache
> >>  shutdown
> >> MIME-Version: 1.0
> >> Content-Type: text/plain; charset=UTF-8
> >> Content-Transfer-Encoding: 8bit
> >> Upon nfsd shutdown any pending DRC cache is freed. DRC cache use is
> >> tracked via a percpu counter. In the current code the percpu counter
> >> is destroyed before. If any pending cache is still present,
> >> percpu_counter_add is called with a percpu counter==NULL. This causes
> >> a kernel crash.
> >> The solution is to destroy the percpu counter after the cache is freed.
> >> Fixes: e567b98ce9a4b (âEURoenfsd: protect concurrent access to nfsd stats countersâEUR)
> >> Signed-off-by: Julian Schroeder <jumaco@amazon.com>
> >> ---
> >>  fs/nfsd/nfscache.c | 2 +-
> >>  1 file changed, 1 insertion(+), 1 deletion(-)
> >> diff --git a/fs/nfsd/nfscache.c b/fs/nfsd/nfscache.c
> >> index 0b3f12aa37ff..7da88bdc0d6c 100644
> >> --- a/fs/nfsd/nfscache.c
> >> +++ b/fs/nfsd/nfscache.c
> >> @@ -206,7 +206,6 @@ void nfsd_reply_cache_shutdown(struct nfsd_net *nn)
> >>         struct svc_cacherep     *rp;
> >>         unsigned int i;
> >>
> >> -       nfsd_reply_cache_stats_destroy(nn);
> >>         unregister_shrinker(&nn->nfsd_reply_cache_shrinker);
> >>
> >>         for (i = 0; i < nn->drc_hashsize; i++) {
> >> @@ -217,6 +216,7 @@ void nfsd_reply_cache_shutdown(struct nfsd_net *nn)
> >>                                                                         rp, nn);
> >>                 }
> >>         }
> >> +       nfsd_reply_cache_stats_destroy(nn);
> >>
> >>         kvfree(nn->drc_hashtbl);
> >>         nn->drc_hashtbl = NULL;
> >> --
> >> 2.32.0
> >>
> 
> >What is the git commit id of this in Linus's tree?
> 
> >And this patch is totally damaged with whitespace and can not be applied
> >at all :(
> 
> >Please fix it up and submit it again.
> 
> >thanks,
> 
> >greg k-h
> 
> The patch made it into linux next. commit fd5e363eac77e.
> I sent the patch to stable again via git send-email (https://www.spinics.net/lists/stable/msg560818.html)

<formletter>

This is not the correct way to submit patches for inclusion in the
stable kernel tree.  Please read:
    https://www.kernel.org/doc/html/latest/process/stable-kernel-rules.html
for how to do this properly.

</formletter>

Re: [PATCH] nfsd: destroy percpu stats counters after reply cache #5.11.0-rc5

[Schroeder, Julian]

>> >> From: Julian Schroeder <jumaco@amazon.com>
>> >> Date: Fri, 20 May 2022 18:33:27 +0000
>> >> Subject: [PATCH] nfsd: destroy percpu stats counters after reply cache
>> >>  shutdown
>> >> MIME-Version: 1.0
>> >> Content-Type: text/plain; charset=UTF-8
>> >> Content-Transfer-Encoding: 8bit
>> >> Upon nfsd shutdown any pending DRC cache is freed. DRC cache use is
>> >> tracked via a percpu counter. In the current code the percpu counter
>> >> is destroyed before. If any pending cache is still present,
>> >> percpu_counter_add is called with a percpu counter==NULL. This causes
>> >> a kernel crash.
>> >> The solution is to destroy the percpu counter after the cache is freed.
>> >> Fixes: e567b98ce9a4b (âEURoenfsd: protect concurrent access to nfsd stats countersâEUR)
>> >> Signed-off-by: Julian Schroeder <jumaco@amazon.com>
>> >> ---
>> >>  fs/nfsd/nfscache.c | 2 +-
>> >>  1 file changed, 1 insertion(+), 1 deletion(-)
>> >> diff --git a/fs/nfsd/nfscache.c b/fs/nfsd/nfscache.c
>> >> index 0b3f12aa37ff..7da88bdc0d6c 100644
>> >> --- a/fs/nfsd/nfscache.c
>> >> +++ b/fs/nfsd/nfscache.c
>> >> @@ -206,7 +206,6 @@ void nfsd_reply_cache_shutdown(struct nfsd_net *nn)
>> >>         struct svc_cacherep     *rp;
>> >>         unsigned int i;
>> >>
>> >> -       nfsd_reply_cache_stats_destroy(nn);
>> >>         unregister_shrinker(&nn->nfsd_reply_cache_shrinker);
>> >>
>> >>         for (i = 0; i < nn->drc_hashsize; i++) {
>> >> @@ -217,6 +216,7 @@ void nfsd_reply_cache_shutdown(struct nfsd_net *nn)
>> >>                                                                         rp, nn);
>> >>                 }
>> >>         }
>> >> +       nfsd_reply_cache_stats_destroy(nn);
>> >>
>> >>         kvfree(nn->drc_hashtbl);
>> >>         nn->drc_hashtbl = NULL;
>> >> --
>> >> 2.32.0
>> >>
>>
>> >What is the git commit id of this in Linus's tree?
>>
>> >And this patch is totally damaged with whitespace and can not be applied
>> >at all :(
>>
>> >Please fix it up and submit it again.
>>
>> >thanks,
>>
>> >greg k-h
>>
>> The patch made it into linux next. commit fd5e363eac77e.
>> I sent the patch to stable again via git send-email (https://www.spinics.net/lists/stable/msg560818.html)
>
><formletter>
>
>This is not the correct way to submit patches for inclusion in the
>stable kernel tree.  Please read:
>    https://www.kernel.org/doc/html/latest/process/stable-kernel-rules.html
>for how to do this properly.
>
></formletter>
Upon nfsd shutdown any pending DRC cache is freed. DRC cache use is
tracked via a percpu counter. In the current code the percpu counter
is destroyed before. If any pending cache is still present,
percpu_counter_add is called with a percpu counter==NULL. This causes
a kernel crash.
The solution is to destroy the percpu counter after the cache is freed.

Fixes: e567b98ce9a4b (“nfsd: protect concurrent access to nfsd stats counters”)
Signed-off-by: Julian Schroeder <jumaco@xxxxxxxxxx>
---
 fs/nfsd/nfscache.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/fs/nfsd/nfscache.c b/fs/nfsd/nfscache.c
index 0b3f12aa37ff..7da88bdc0d6c 100644
--- a/fs/nfsd/nfscache.c
+++ b/fs/nfsd/nfscache.c
@@ -206,7 +206,6 @@ void nfsd_reply_cache_shutdown(struct nfsd_net *nn)
 	struct svc_cacherep	*rp;
 	unsigned int i;
 
-	nfsd_reply_cache_stats_destroy(nn);
 	unregister_shrinker(&nn->nfsd_reply_cache_shrinker);
 
 	for (i = 0; i < nn->drc_hashsize; i++) {
@@ -217,6 +216,7 @@ void nfsd_reply_cache_shutdown(struct nfsd_net *nn)
 									rp, nn);
 		}
 	}
+	nfsd_reply_cache_stats_destroy(nn);
 
 	kvfree(nn->drc_hashtbl);
 	nn->drc_hashtbl = NULL;
-- 
2.32.0

Re: [PATCH] nfsd: destroy percpu stats counters after reply cache #5.11.0-rc5

[Greg KH]

On Sun, May 29, 2022 at 12:57:59PM +0000, Schroeder, Julian wrote:
> >> >> From: Julian Schroeder <jumaco@amazon.com>
> >> >> Date: Fri, 20 May 2022 18:33:27 +0000
> >> >> Subject: [PATCH] nfsd: destroy percpu stats counters after reply cache
> >> >>  shutdown
> >> >> MIME-Version: 1.0
> >> >> Content-Type: text/plain; charset=UTF-8
> >> >> Content-Transfer-Encoding: 8bit
> >> >> Upon nfsd shutdown any pending DRC cache is freed. DRC cache use is
> >> >> tracked via a percpu counter. In the current code the percpu counter
> >> >> is destroyed before. If any pending cache is still present,
> >> >> percpu_counter_add is called with a percpu counter==NULL. This causes
> >> >> a kernel crash.
> >> >> The solution is to destroy the percpu counter after the cache is freed.
> >> >> Fixes: e567b98ce9a4b (âEURoenfsd: protect concurrent access to nfsd stats countersâEUR)
> >> >> Signed-off-by: Julian Schroeder <jumaco@amazon.com>
> >> >> ---
> >> >>  fs/nfsd/nfscache.c | 2 +-
> >> >>  1 file changed, 1 insertion(+), 1 deletion(-)
> >> >> diff --git a/fs/nfsd/nfscache.c b/fs/nfsd/nfscache.c
> >> >> index 0b3f12aa37ff..7da88bdc0d6c 100644
> >> >> --- a/fs/nfsd/nfscache.c
> >> >> +++ b/fs/nfsd/nfscache.c
> >> >> @@ -206,7 +206,6 @@ void nfsd_reply_cache_shutdown(struct nfsd_net *nn)
> >> >>         struct svc_cacherep     *rp;
> >> >>         unsigned int i;
> >> >>
> >> >> -       nfsd_reply_cache_stats_destroy(nn);
> >> >>         unregister_shrinker(&nn->nfsd_reply_cache_shrinker);
> >> >>
> >> >>         for (i = 0; i < nn->drc_hashsize; i++) {
> >> >> @@ -217,6 +216,7 @@ void nfsd_reply_cache_shutdown(struct nfsd_net *nn)
> >> >>                                                                         rp, nn);
> >> >>                 }
> >> >>         }
> >> >> +       nfsd_reply_cache_stats_destroy(nn);
> >> >>
> >> >>         kvfree(nn->drc_hashtbl);
> >> >>         nn->drc_hashtbl = NULL;
> >> >> --
> >> >> 2.32.0
> >> >>
> >>
> >> >What is the git commit id of this in Linus's tree?
> >>
> >> >And this patch is totally damaged with whitespace and can not be applied
> >> >at all :(
> >>
> >> >Please fix it up and submit it again.
> >>
> >> >thanks,
> >>
> >> >greg k-h
> >>
> >> The patch made it into linux next. commit fd5e363eac77e.
> >> I sent the patch to stable again via git send-email (https://www.spinics.net/lists/stable/msg560818.html)
> >
> ><formletter>
> >
> >This is not the correct way to submit patches for inclusion in the
> >stable kernel tree.  Please read:
> >    https://www.kernel.org/doc/html/latest/process/stable-kernel-rules.html
> >for how to do this properly.
> >
> ></formletter>
> Upon nfsd shutdown any pending DRC cache is freed. DRC cache use is
> tracked via a percpu counter. In the current code the percpu counter
> is destroyed before. If any pending cache is still present,
> percpu_counter_add is called with a percpu counter==NULL. This causes
> a kernel crash.
> The solution is to destroy the percpu counter after the cache is freed.
> 
> Fixes: e567b98ce9a4b (“nfsd: protect concurrent access to nfsd stats counters”)
> Signed-off-by: Julian Schroeder <jumaco@xxxxxxxxxx>

That is a very odd email address :(

Again, as the document says, just wait until a commit is in Linus's tree
and then send us the git commit id.

thanks,

greg k-h