linux-kernel - Re: [PATCH] PCI/DOE: Fix destroy_work_on

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: <20230727202149.GA793572@bhelgaas>
Date:   Thu, 27 Jul 2023 15:21:49 -0500
From:   Bjorn Helgaas <helgaas@...nel.org>
To:     Ira Weiny <ira.weiny@...el.com>
Cc:     Bjorn Helgaas <bhelgaas@...gle.com>,
        Dan Williams <dan.j.williams@...el.com>,
        Jonathan Cameron <Jonathan.Cameron@...wei.com>,
        Lukas Wunner <lukas@...ner.de>,
        Davidlohr Bueso <dave@...olabs.net>, linux-pci@...r.kernel.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH] PCI/DOE: Fix destroy_work_on_stack() race

On Wed, Jul 26, 2023 at 11:29:42AM -0700, Ira Weiny wrote:
> The following debug object splat was observed in testing.
> 
>   [   14.061937] ------------[ cut here ]------------
>   [   14.063899] ODEBUG: free active (active state 0) object: 0000000097d23782 object type: work_struct hint: doe_statemachine_work+0x0/0x510
>   [   14.067480] WARNING: CPU: 1 PID: 71 at lib/debugobjects.c:514 debug_print_object+0x7d/0xb0
>   ...
>   [   14.080951] Workqueue: pci 0000:36:00.0 DOE [1 doe_statemachine_work
>   [   14.083485] RIP: 0010:debug_print_object+0x7d/0xb0
>   ...
>   [   14.116231] Call Trace:
>   [   14.117652]  <TASK>
>   [   14.118958]  ? debug_print_object+0x7d/0xb0
>   [   14.120782]  ? __warn+0x7d/0x130
>   [   14.122399]  ? debug_print_object+0x7d/0xb0
>   [   14.123746]  ? report_bug+0x18d/0x1c0
>   [   14.125025]  ? handle_bug+0x3c/0x80
>   [   14.126506]  ? exc_invalid_op+0x13/0x60
>   [   14.127796]  ? asm_exc_invalid_op+0x16/0x20
>   [   14.129380]  ? debug_print_object+0x7d/0xb0
>   [   14.130688]  ? debug_print_object+0x7d/0xb0
>   [   14.131997]  ? __pfx_doe_statemachine_work+0x10/0x10
>   [   14.133597]  debug_object_free.part.0+0x11b/0x150
>   [   14.134940]  doe_statemachine_work+0x45e/0x510
>   [   14.136348]  process_one_work+0x1d4/0x3c0
>   ...
>   [   14.161484]  </TASK>
>   [   14.162434] ---[ end trace 0000000000000000 ]---
> 
> This occurs because destroy_work_on_stack() was called after signaling
> the completion in the calling thread.  This creates a race between
> destroy_work_on_stack() and the task->work struct going of scope in the
> pci_doe().
> 
> Signal the work complete after destroying the work struct.  This is safe
> because signal_task_complete() is the final thing the work item does and
> the workqueue code is careful not to access the work struct after.
> 
> Fixes: abf04be0e707 ("PCI/DOE: Fix memory leak with CONFIG_DEBUG_OBJECTS=y")
> Cc: Lukas Wunner <lukas@...ner.de>
> Signed-off-by: Ira Weiny <ira.weiny@...el.com>

Thanks, applied to pci/misc with Lukas' reviewed-by and Dan's ack for
v6.6.  I edited out the timestamps and some of the call trace from the
splat because they didn't seem relevant.

> ---
>  drivers/pci/doe.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/pci/doe.c b/drivers/pci/doe.c
> index 1b97a5ab71a9..e3aab5edaf70 100644
> --- a/drivers/pci/doe.c
> +++ b/drivers/pci/doe.c
> @@ -293,8 +293,8 @@ static int pci_doe_recv_resp(struct pci_doe_mb *doe_mb, struct pci_doe_task *tas
>  static void signal_task_complete(struct pci_doe_task *task, int rv)
>  {
>  	task->rv = rv;
> -	task->complete(task);
>  	destroy_work_on_stack(&task->work);
> +	task->complete(task);
>  }
>  
>  static void signal_task_abort(struct pci_doe_task *task, int rv)
> 
> ---
> base-commit: 20ea1e7d13c1b544fe67c4a8dc3943bb1ab33e6f
> change-id: 20230726-doe-fix-f57943f9ea82
> 
> Best regards,
> -- 
> Ira Weiny <ira.weiny@...el.com>
>