Re: [PATCH] tee: fix tee_ioctl_object_invoke_arg padding

From: Harshal Dev <harshal.dev@oss.qualcomm.com>
To: Sumit Garg <sumit.garg@kernel.org>,
	Amirreza Zarrabi <amirreza.zarrabi@oss.qualcomm.com>,
	Arnd Bergmann <arnd@arndb.de>,
	Jens Wiklander <jens.wiklander@linaro.org>
Cc: Arnd Bergmann <arnd@kernel.org>, Paul Walmsley <pjw@kernel.org>,
	Palmer Dabbelt <palmer@dabbelt.com>,
	Albert Ou <aou@eecs.berkeley.edu>,
	Sumit Garg <sumit.garg@oss.qualcomm.com>,
	Alexandre Ghiti <alex@ghiti.fr>,
	Etienne Carriere <etienne.carriere@foss.st.com>,
	Randy Dunlap <rdunlap@infradead.org>,
	op-tee@lists.trustedfirmware.org, linux-kernel@vger.kernel.org,
	linux-riscv@lists.infradead.org
Subject: Re: [PATCH] tee: fix tee_ioctl_object_invoke_arg padding
Date: Mon, 8 Dec 2025 18:24:15 +0530	[thread overview]
Message-ID: <2c0d09b7-0d8c-4bf2-a1bd-1dc4b58cd0a8@oss.qualcomm.com> (raw)
In-Reply-To: <aTbCqyOuprPOyU4X@sumit-X1>

On 12/8/2025 5:50 PM, Sumit Garg via OP-TEE wrote:
> On Mon, Dec 08, 2025 at 04:24:17PM +1100, Amirreza Zarrabi wrote:
>> Hi,
>>
>> On 12/5/2025 12:27 AM, Jens Wiklander wrote:
>>> Hi,
>>>
>>> On Thu, Dec 4, 2025 at 11:17 AM Arnd Bergmann <arnd@kernel.org> wrote:
>>>>
>>>> From: Arnd Bergmann <arnd@arndb.de>
>>>>
>>>> The tee_ioctl_object_invoke_arg structure has padding on some
>>>> architectures but not on x86-32 and a few others:
>>>>
>>>> include/linux/tee.h:474:32: error: padding struct to align 'params' [-Werror=padded]
>>>>
>>>> I expect that all current users of this are on architectures that do
>>>> have implicit padding here (arm64, arm, x86, riscv), so make the padding
>>>> explicit in order to avoid surprises if this later gets used elsewhere.
>>>>
>>>> Fixes: d5b8b0fa1775 ("tee: add TEE_IOCTL_PARAM_ATTR_TYPE_OBJREF")
>>>> Signed-off-by: Arnd Bergmann <arnd@arndb.de>
>>>> ---
>>>> The new interface showed up in 6.18, but I only came across this after
>>>> that was released. Changing it now is technically an ABI change on
>>>> architectures with unusual padding rules, so please consider carefully
>>>> whether we want to do it this way or not.
>>>>
>>>> Working around the ABI differences without an ABI change is possible,
>>>> but adds a lot of complexity for compat handling.
>>>
>>> This is currently only used by the recently introduced qcomtee backend
>>> driver. So it's only used on a few arm64 Qualcomm platforms right now.
>>>
>>> I think we should take this patch, but let's hear what others think.
> 
> Yeah since it's not an ABI issue on arm64 platforms where QTEE runs, so:
> 
> Reviewed-by: Sumit Garg <sumit.garg@oss.qualcomm.com>
> 
>>>
>>> Thanks,
>>> Jens
>>>
>>
>> I agree. We should take this patch. As noted, there are not many
>> clients relying on it yet, so updating the userspace should
>> be straightforward.
> 
> You should rather test without any userspace library update to test it's
> not an ABI issue. Just for correctness sake, you can update the library
> too.
> 

I'll take the time to test it at some point this week both with and without updating
the library ABI.

Regards,
Harshal

> -Sumit
> 
> [...]

_______________________________________________
linux-riscv mailing list
linux-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-riscv