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: <47aa934d-bdad-4c76-9420-ce65950f67d1@linaro.org>
Date: Wed, 28 Aug 2024 01:07:49 +0100
From: Bryan O'Donoghue <bryan.odonoghue@...aro.org>
To: Vladimir Zapolskiy <vladimir.zapolskiy@...aro.org>,
 Depeng Shao <quic_depengs@...cinc.com>, rfoss@...nel.org,
 todor.too@...il.com, mchehab@...nel.org, robh@...nel.org,
 krzk+dt@...nel.org, conor+dt@...nel.org
Cc: linux-arm-msm@...r.kernel.org, linux-media@...r.kernel.org,
 devicetree@...r.kernel.org, linux-kernel@...r.kernel.org, kernel@...cinc.com
Subject: Re: [PATCH 09/13] media: qcom: camss: vfe: Move common code into vfe
 core

On 24/08/2024 14:06, Vladimir Zapolskiy wrote:
>> Right so generally speaking I don't believe we should have any null
>> function pointers.
>>
>> We just mandate that to be comitted, an impelmentation must provide a
>> dummy but, in this case when do we ever want a dummy function anyway
>> surely enable_irq() is a fundamental operation that is core to the logic.
> 
> Why? What could be a justification here?

a) In principle I don't believe we should be adding an interface that 
requires an exception in its first go.

b) I don't think its a good idea to proliferate patterns like this 
throughout the code

if (vfe->res->hw_ops->enable_irq)
     vfe->res->hw_ops->enable_irq();

too->many->indirection->calls();

easily dealt with but "looks wrong"

https://www.joelonsoftware.com/2005/05/11/making-wrong-code-look-wrong/

So if we are adding callbacks, the exception should be stubbing to an 
empty function because 9/10 of implementations find the interface useful.

> It might be extremely confusing to see in the code that some not ever
> requested interrupts are enabled/disabled, and then to discover that just
> some stubs around VFE interrupts are added. And it's the case especially
> in this new vfe_enable_v2() function, which I believe is intended for
> CAMSS support on new platforms.
> 
> What's worse, since these VFE interrupts are not needed on the modern
> platforms, it will require to add a proposed dummy "return 0" function
> into any CAMSS support for new platforms forever. I believe it'd be better
> to clearly say that it's a legacy to have an obligatory support of VFE
> interrupts.

I think enable_irq();/disable_irq(); should be handled from wm_start() 
and wm_stop() for each VFE so that 480 and after can have shared 
top-level logic.

However VFE can raise several error cases which TBH we should probably 
be capturing somehow, if not in IRQ then in debugfs and/or trace

https://git.codelinaro.org/bryan.odonoghue/kernel/-/commit/cd88d924eb55f5dfeb2283e6e0eef37d5bd4c1c4

@Depeng can you move your enable_irq(); to camss-vfe-480.c::wm_start()

just after we switch on a WM

https://git.codelinaro.org/bryan.odonoghue/kernel/-/blob/2ea8172164e2b12a629cf3d939edac9a0f7a9368/drivers/media/platform/qcom/camss/camss-vfe-480.c#L127

as well as a disable_irq() to camss-vfe-480.c::wm_stop(); @ the top of 
the function ?

https://git.codelinaro.org/bryan.odonoghue/kernel/-/blob/2ea8172164e2b12a629cf3d939edac9a0f7a9368/drivers/media/platform/qcom/camss/camss-vfe-480.c#L131

disable_irq() should also collapse the TOP irq when stream_count == 0

The stream_count variable should ensure the TOP irq value only gets 
switched on/off once while also removing the special case from your 
proposed amalgamated code.

---
bod

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ