Re: [PATCH 17/17] mmc: pass queue_limits to blk_mq_alloc_disk

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 




On 27/06/2024 12:19, Christoph Hellwig wrote:
On Thu, Jun 27, 2024 at 10:58:58AM +0100, Jon Hunter wrote:

On 27/06/2024 10:49, Christoph Hellwig wrote:
On Thu, Jun 27, 2024 at 10:43:24AM +0100, Jon Hunter wrote:
We have just noticed that since Linux v6.9 was released, that if we
build the kernel with 64kB MMU pages, then we see the following WARNING
and probe failure ...

The old code upgraded the limits to the PAGE_SIZE for this case after
issunig a warning.  Your driver probably incorrectly advertised the
lower max_segment_size.  Try setting it to 64k.  I would have sent you
a patch for that, but I can't see what mmc host driver you are using.


We are using the sdhci-tegra.c driver. I don't see it set in there, but I
see references to max_seg_size in the main sdhci.c driver.

sdhci-tegra.c sets SDHCI_QUIRK_BROKEN_ADMA_ZEROLEN_DESC for various
devices, I guess you have one od those?  This sounds like the device
might not actually need it, but I'll need help from the SDHCI and
tegra maintainers here.


I have been testing on both Tegra194 and Tegra234. Both of these set the above quirk. This would explain why the max_segment_size is rounded down to 65024 in the mmc_alloc_disk() function.

We can check if this is needed but if it is needed then it is not clear if/how this can be fixed?

Thanks
Jon

--
nvpublic




[Index of Archives]     [ARM Kernel]     [Linux ARM]     [Linux ARM MSM]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]

  Powered by Linux