Netdev Archive on
help / color / mirror / Atom feed
From: Wen Gu <>
To: Karsten Graul <>,,
Subject: Re: [PATCH net] net/smc: Avoid setting clcsock options after clcsock released
Date: Wed, 12 Jan 2022 00:34:16 +0800	[thread overview]
Message-ID: <> (raw)
In-Reply-To: <>

Thanks for your reply.

On 2022/1/11 6:03 pm, Karsten Graul wrote:
> On 10/01/2022 10:38, Wen Gu wrote:
>> We encountered a crash in smc_setsockopt() and it is caused by
>> accessing smc->clcsock after clcsock was released.
> In the switch() the function smc_switch_to_fallback() might be called which also
> accesses smc->clcsock without further checking. This should also be protected then?
> Also from all callers of smc_switch_to_fallback() ?
> There are more uses of smc->clcsock (e.g. smc_bind(), ...), so why does this problem
> happen in setsockopt() for you only? I suspect it depends on the test case.

Yes, it depends on the test case. The crash described here only happens one time when
I run a stress test of nginx/wrk, accompanied with frequent RNIC up/down operations.

Considering accessing smc->clcsock after its release is an uncommon, low probability
issue and only happens in setsockopt() in my test, I choce an simple way to fix it, using
the existing clcsock_release_lock, and only check in smc_setsockopt() and smc_getsockopt().

> I wonder if it makes sense to check and protect smc->clcsock at all places in the code where
> it is used... as of now we had no such races like you encountered. But I see that in theory
> this problem could also happen in other code areas.

But inspired by your questions, I think maybe we should treat the race as a general problem?
Do you think it is necessary to find all the potential race related to the clcsock release and
fix them in a unified approach? like define smc->clcsock as RCU pointer, hold rcu read lock
before accessing smc->clcsock and call synchronize_rcu() before resetting smc->clcsock? just a rough idea :)

Or we should decide it later, do some more tests to see the probability of recurrence of this problem?

Wen Gu

  reply	other threads:[~2022-01-11 16:34 UTC|newest]

Thread overview: 10+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-01-10  9:38 Wen Gu
2022-01-11 10:03 ` Karsten Graul
2022-01-11 16:34   ` Wen Gu [this message]
2022-01-12  9:38     ` Karsten Graul
2022-01-13  8:23       ` Wen Gu
2022-01-13 15:15       ` Wen Gu
2022-01-11 18:14 ` Jakub Kicinski
2022-01-12  3:32   ` Wen Gu
2022-01-12  7:11 `
2022-01-12  8:16   ` Wen Gu

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \ \ \ \ \ \ \ \ \
    --subject='Re: [PATCH net] net/smc: Avoid setting clcsock options after clcsock released' \

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).