From: Marcin Slusarz <marcin.slusarz@gmail.com>
To: Kulikov Vasiliy <segooon@gmail.com>
Cc: kernel-janitors@vger.kernel.org,
Richard Henderson <rth@twiddle.net>,
Ivan Kokshaysky <ink@jurassic.park.msu.ru>,
Matt Turner <mattst88@gmail.com>,
linux-alpha@vger.kernel.org, linux-kernel@vger.kernel.org,
Andrew Morton <akpm@linux-foundation.org>
Subject: Re: [PATCH 01/15] alpha: srmcons: localize spin_lock section
Date: Fri, 16 Jul 2010 20:30:55 +0000 [thread overview]
Message-ID: <20100716203055.GA2731@joi.lan> (raw)
In-Reply-To: <1279296747-24783-1-git-send-email-segooon@gmail.com>
On Fri, Jul 16, 2010 at 08:12:27PM +0400, Kulikov Vasiliy wrote:
> Scale down spin_lock_irqsave/spin_unlock_irqrestore exactly to area
> where it is needed. Also it makes static checkers happy as code checks
> kmalloc() result value exactly after call.
>
> Signed-off-by: Kulikov Vasiliy <segooon@gmail.com>
> ---
> arch/alpha/kernel/srmcons.c | 6 ++----
> 1 files changed, 2 insertions(+), 4 deletions(-)
>
> diff --git a/arch/alpha/kernel/srmcons.c b/arch/alpha/kernel/srmcons.c
> index 783f4e5..6b04005 100644
> --- a/arch/alpha/kernel/srmcons.c
> +++ b/arch/alpha/kernel/srmcons.c
> @@ -165,17 +165,15 @@ srmcons_get_private_struct(struct srmcons_private **ps)
>
> if (srmconsp = NULL) {
> srmconsp = kmalloc(sizeof(*srmconsp), GFP_KERNEL);
> - spin_lock_irqsave(&srmconsp_lock, flags);
> -
> if (srmconsp = NULL)
> retval = -ENOMEM;
> else {
> + spin_lock_irqsave(&srmconsp_lock, flags);
> srmconsp->tty = NULL;
> spin_lock_init(&srmconsp->lock);
> init_timer(&srmconsp->timer);
> + spin_unlock_irqrestore(&srmconsp_lock, flags);
> }
> -
> - spin_unlock_irqrestore(&srmconsp_lock, flags);
> }
>
> *ps = srmconsp;
> --
Locking in this function seems to be broken. If we are unlucky:
- two cpus will allocate, initialize and return two different pointers
- two cpus will allocate, but both initialize only one (two times!) and return
the same pointer (so one of them will be leaked) - 2 initializations may lead
to double unlock later, etc
- ...
Something like patch below is needed. I don't have alpha cross compiler so
I can't even compile it...
---
From: Marcin Slusarz <marcin.slusarz@gmail.com>
Subject: [PATCH] alpha: fix races in srmcons_get_private_struct
---
arch/alpha/kernel/srmcons.c | 38 ++++++++++++++++++++++++--------------
1 files changed, 24 insertions(+), 14 deletions(-)
diff --git a/arch/alpha/kernel/srmcons.c b/arch/alpha/kernel/srmcons.c
index 783f4e5..531ac99 100644
--- a/arch/alpha/kernel/srmcons.c
+++ b/arch/alpha/kernel/srmcons.c
@@ -161,25 +161,35 @@ srmcons_get_private_struct(struct srmcons_private **ps)
static struct srmcons_private *srmconsp = NULL;
static DEFINE_SPINLOCK(srmconsp_lock);
unsigned long flags;
- int retval = 0;
+ struct srmcons_private *tmp;
- if (srmconsp = NULL) {
- srmconsp = kmalloc(sizeof(*srmconsp), GFP_KERNEL);
- spin_lock_irqsave(&srmconsp_lock, flags);
-
- if (srmconsp = NULL)
- retval = -ENOMEM;
- else {
- srmconsp->tty = NULL;
- spin_lock_init(&srmconsp->lock);
- init_timer(&srmconsp->timer);
- }
+ if (srmconsp) {
+ *ps = srmconsp;
+ return 0;
+ }
- spin_unlock_irqrestore(&srmconsp_lock, flags);
+ tmp = kmalloc(sizeof(*srmconsp), GFP_KERNEL);
+ if (!tmp) {
+ *ps = NULL;
+ return -ENOMEM;
}
+ spin_lock_irqsave(&srmconsp_lock, flags);
+
+ /* recheck under lock to not race with another cpu */
+ if (srmconsp = NULL) {
+ tmp->tty = NULL;
+ spin_lock_init(&tmp->lock);
+ init_timer(&tmp->timer);
+
+ srmconsp = tmp;
+ } else
+ kfree(tmp);
+
+ spin_unlock_irqrestore(&srmconsp_lock, flags);
+
*ps = srmconsp;
- return retval;
+ return 0;
}
static int
--
1.7.1
prev parent reply other threads:[~2010-07-16 20:30 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-07-16 16:12 [PATCH 01/15] alpha: srmcons: localize spin_lock section Kulikov Vasiliy
2010-07-16 20:30 ` Marcin Slusarz [this message]
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:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20100716203055.GA2731@joi.lan \
--to=marcin.slusarz@gmail.com \
--cc=akpm@linux-foundation.org \
--cc=ink@jurassic.park.msu.ru \
--cc=kernel-janitors@vger.kernel.org \
--cc=linux-alpha@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=mattst88@gmail.com \
--cc=rth@twiddle.net \
--cc=segooon@gmail.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
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).