linux-kernel - Re: [PATCH] soc: qcom: icc-bwmon: Don't ignore return values of regmap functions

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 for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Tue, 20 Jun 2023 20:15:05 +0200
From:   Krzysztof Kozlowski <krzysztof.kozlowski@...aro.org>
To:     Bjorn Andersson <andersson@...nel.org>
Cc:     Konrad Dybcio <konrad.dybcio@...aro.org>,
        Andy Gross <agross@...nel.org>,
        Marijn Suijten <marijn.suijten@...ainline.org>,
        linux-arm-msm@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] soc: qcom: icc-bwmon: Don't ignore return values of
 regmap functions

On 20/06/2023 20:14, Bjorn Andersson wrote:
> On Thu, Jun 15, 2023 at 11:26:13PM +0200, Krzysztof Kozlowski wrote:
>> On 15/06/2023 23:12, Konrad Dybcio wrote:
>>> As it turns out, not all regmap accesses succeed. Not knowing this is
>>> particularly suboptimal when there's a breaking change to the regmap
>>> APIs. Monitor the return values of regmap_ calls and propagate errors,
>>> should any occur.
>>>
>>> To keep any level of readability in bwmon_enable(), add some comments
>>> to separate the logical blocks.
>>>
>>> Signed-off-by: Konrad Dybcio <konrad.dybcio@...aro.org>
>>
>> Nice coincidence, I just had some talks with a friend about uselessness
>> (IMHO) of regmap MMIO return status checks.
>>
>> Sorry, for me most of this makes the code difficult to read for no gain.
>> Errors are not real. This is some artificial problem. Solving it makes
>> code less maintainable.
>>
>> If we used here readl/writel, you would not add any checks, right? Then
>> don't add for regmap mmio.
>>
> 
> I agree, the mmio regmap interface should only fail because of bugs or
> things are misconfigured. Would be nice to capture that in a WARN_ON()
> or something...
> 

One choice could be to have for entire functions doing reads/writes:

	ret = 0;
	ret != regmap_write();
	ret != regmap_write();
	ret != regmap_write();
	return ret;

and handle this in the caller somehow. I don't think that aborting such
chain early, just because regmap mmio failures, makes sense.

Best regards,
Krzysztof