Re: Linux 6.16-rc1

[Pratyush Yadav <pratyush@kernel.org>]

Hi Linus,

On Fri, Jun 20 2025, Linus Torvalds wrote:

> On Tue, 10 Jun 2025 at 18:25, Guenter Roeck <linux@roeck-us.net> wrote:
>>
>> The test failures are all due to commit 0aa7b390fc40 ("mtd: core: always
>> create master device") which breaks mtd partitioning.
>
> Ok, I just merged the pull request that reverted that one.
>
>> One side note: Various qemu machines configured to use Macronix flash chips
>> are no longer able to access the chips. This is primarily due to commit
>> 947c86e481a0 ("mtd: spi-nor: macronix: Drop the redundant flash info
>> fields"). After that change, SFDP support for affected chips is mandatory,
>> and qemu does not or not correctly support that. This affects quanta-gsj,
>> kudo-bmc, ast2600-evb, and supermicro-x11spi-bmc machines (and possibly
>> others).
>
> .. but this issue presumably remains, unless there's been some subtler
> indirect fix that I didn't realize.
>
> Miguel, that one came through you too. Comments? I do think that
> running things in qemu is likely important for a lot of these things
> that don't actually have a ton of hardware that is easily automated...

There isn't a fix AFAIK. From the thread [0], there seemed to be a
conclusion that it was a qemu problem and not directly a problem on the
driver side. So I dropped this down my list of priorities of things to
look into.

I don't have much idea of how people use qemu for testing, but since you
say this is important for testing workloads, I can take a deeper dive
next week and have an answer by -rc4.

[0] https://lore.kernel.org/linux-mtd/8d4bd876-3571-46e5-857a-948e58b21c5b@roeck-us.net/

-- 
Regards,
Pratyush Yadav