linux-kernel - Re: [PATCH v1 0/4] Add tracing for SDHCI register access

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]

Date:   Thu, 18 Apr 2019 09:24:16 +0300
From:   Adrian Hunter <adrian.hunter@...el.com>
To:     Raul Rangel <rrangel@...omium.org>
Cc:     linux-trace-devel@...r.kernel.org, linux-mmc@...r.kernel.org,
        djkurtz@...omium.org, zwisler@...omium.org,
        Ludovic Barre <ludovic.barre@...com>,
        Steven Rostedt <rostedt@...dmis.org>,
        Jisheng Zhang <Jisheng.Zhang@...aptics.com>,
        Masahiro Yamada <yamada.masahiro@...ionext.com>,
        Faiz Abbas <faiz_abbas@...com>, linux-kernel@...r.kernel.org,
        Oleksij Rempel <linux@...pel-privat.de>,
        Liming Sun <lsun@...lanox.com>,
        Wolfram Sang <wsa+renesas@...g-engineering.com>,
        Ingo Molnar <mingo@...hat.com>,
        Prabu Thangamuthu <Prabu.T@...opsys.com>,
        Chunyan Zhang <chunyan.zhang@...soc.com>,
        Ulf Hansson <ulf.hansson@...aro.org>
Subject: Re: [PATCH v1 0/4] Add tracing for SDHCI register access

On 12/04/19 5:50 PM, Raul Rangel wrote:
> On Fri, Apr 12, 2019 at 09:26:44AM +0300, Adrian Hunter wrote:
>> On 12/04/19 1:08 AM, Raul E Rangel wrote:
>>> I was debugging a SDHC hardware bug and got tired of having to
>>> translate the register values by hand. This patch set makes it so all
>>> SDHC register read and write operations can be traced and easily read by
>>> a human.
>>
>> While this might be useful for people unfamiliar with SDHCI, I am not sure
>> it should be in the upstream kernel.
> Can you help me understand your hesitation? Would you prefer removing the
> pretty printing? Or would you prefer not having any trace events at all?

It is just cost/benefit.  I am not sure the cost of maintaining it is worth
the benefit of having it upstream.

> The xhci driver has a bunch of pretty print trace events that make it
> invaluable when debugging.
> https://github.com/torvalds/linux/blob/d7563ca5bfca53398e100eb74345c5d3ef06bf9d/drivers/usb/host/xhci.h#L2160
> 
>> Also, it doesn't seem ideal for every
>> driver to add its own plumbing for such a feature.
> What do you mean by every driver having to add it's own plumbing? Any
> driver that uses sdhci_readX or sdhci_writeX get the functionality for
> free.

I/O memory mappings are recorded for devices. e.g. /proc/iomem so it could
be possible to add tracing for all I/O memory accesses via readx/writex etc,
configurable / filterable by device id.