lists.openwall.net   lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  linux-cve-announce  PHC 
Open Source and information security mailing list archives
 
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <68f3e5f8-895e-416b-88cf-284a263bd954@stanley.mountain>
Date: Thu, 3 Oct 2024 17:31:53 +0300
From: Dan Carpenter <dan.carpenter@...aro.org>
To: Jens Axboe <axboe@...nel.dk>
Cc: Waiman Long <longman@...hat.com>, Yu Kuai <yukuai3@...wei.com>,
	Tejun Heo <tj@...nel.org>, Josef Bacik <josef@...icpanda.com>,
	cgroups@...r.kernel.org, linux-block@...r.kernel.org,
	linux-kernel@...r.kernel.org, kernel-janitors@...r.kernel.org,
	Christoph Hellwig <hch@....de>
Subject: Re: [PATCH v2] blk_iocost: remove some duplicate irq disable/enables

On Thu, Oct 03, 2024 at 07:21:25AM -0600, Jens Axboe wrote:
> On 10/3/24 6:03 AM, Dan Carpenter wrote:
> >   3117                                  ioc_now(iocg->ioc, &now);
> >   3118                                  weight_updated(iocg, &now);
> >   3119                                  spin_unlock(&iocg->ioc->lock);
> >   3120                          }
> >   3121                  }
> >   3122                  spin_unlock_irq(&blkcg->lock);
> >   3123  
> >   3124                  return nbytes;
> >   3125          }
> >   3126  
> >   3127          blkg_conf_init(&ctx, buf);
> >   3128  
> >   3129          ret = blkg_conf_prep(blkcg, &blkcg_policy_iocost, &ctx);
> >   3130          if (ret)
> >   3131                  goto err;
> >   3132  
> >   3133          iocg = blkg_to_iocg(ctx.blkg);
> >   3134  
> >   3135          if (!strncmp(ctx.body, "default", 7)) {
> >   3136                  v = 0;
> >   3137          } else {
> >   3138                  if (!sscanf(ctx.body, "%u", &v))
> >   3139                          goto einval;
> >   3140                  if (v < CGROUP_WEIGHT_MIN || v > CGROUP_WEIGHT_MAX)
> >   3141                          goto einval;
> >   3142          }
> >   3143  
> >   3144          spin_lock(&iocg->ioc->lock);
> > 
> > But why is this not spin_lock_irq()?  I haven't analyzed this so maybe it's
> > fine.
> 
> That's a bug.
> 

I could obviously write this patch but I feel stupid writing the commit message.
My level of understanding is Monkey See Monkey do.  Could you take care of this?

So somewhere we're taking a lock in the IRQ handler and this can lead to a
deadlock? I thought this would have been caught by lockdep?

regards,
dan carpenter

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ