linux-kernel - Re: [PATCH v2 2/2] bus: mhi: host: Take irqsave lock after TRE is generated

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: <10f05b5a-8046-4884-b375-007a14248e00@quicinc.com>
Date:   Mon, 16 Oct 2023 16:49:08 +0800
From:   Qiang Yu <quic_qianyu@...cinc.com>
To:     Jeffrey Hugo <quic_jhugo@...cinc.com>, <mani@...nel.org>
CC:     <mhi@...ts.linux.dev>, <linux-arm-msm@...r.kernel.org>,
        <linux-kernel@...r.kernel.org>, <quic_cang@...cinc.com>,
        <quic_mrana@...cinc.com>, Hemant Kumar <quic_hemantk@...cinc.com>,
        "Lazarus Motha" <quic_lmotha@...cinc.com>
Subject: Re: [PATCH v2 2/2] bus: mhi: host: Take irqsave lock after TRE is
 generated


On 9/29/2023 11:25 PM, Jeffrey Hugo wrote:
> On 9/24/2023 10:08 PM, Qiang Yu wrote:
>>
>> On 9/22/2023 10:50 PM, Jeffrey Hugo wrote:
>>> On 9/13/2023 2:47 AM, Qiang Yu wrote:
>>>> From: Hemant Kumar <quic_hemantk@...cinc.com>
>>>>
>>>> Take irqsave lock after TRE is generated to avoid deadlock due to core
>>>> getting interrupts enabled as local_bh_enable must not be called with
>>>> irqs disabled based on upstream patch.
>>>
>>> Where is local_bh_enable() being called?  What patch?  What is 
>>> upstream of the codebase you submitted this to?  Why is it safe to 
>>> call mhi_gen_tre() without the lock?
>>
>> This patch is to fix the issue included by  "[PATCH v2 1/2] bus: mhi: 
>> host: Add spinlock to protect WP access when queueing TREs". In that 
>> patch, we add write_lock_bh/write_unlock_bh in mhi_gen_tre().
>>
>> However, before mhi_gen_tre() is invoked, mhi_cntrl->pm_lock is 
>> getted, line 1125, and it is a spin lock. So it becomes we want to 
>> get and release bh lock after spin lock. __local_bh_enable_ip is 
>> called as part of write_unlock_bh
>>
>> and local_bh_enable. When CONFIG_TRACE_IRQFLAGS is enabled, irq will 
>> be enabled once __local_bh_enable_ip is called. The commit message is 
>> not clear and confusing, will change it in [patch v3].
>>
>
> In addition to clarifying the commit message, I recommend looking at 
> adding this to the other patch.  It seems very odd to review a series 
> where one patch introduces a known issue, and a following patch 
> corrects that issue.  It would be better to not introduce the issue in 
> the first place.
OK, will squash two patches into one patch after we achieve an agreement.