Re: [6.12.y] WARNING: CPU: 0 PID: 681 at block/blk-mq-cpumap.c:90 blk_mq_map_hw_queues+0xcf/0xe0

[Greg KH <gregkh@linuxfoundation.org>]

On Mon, May 05, 2025 at 03:40:51PM +0200, Daniel Wagner wrote:
> On Mon, May 05, 2025 at 03:28:22PM +0200, Greg KH wrote:
> > On Mon, May 05, 2025 at 01:36:52PM +0200, Jack Wang wrote:
> > > From: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
> > > 
> > > In linux-6.12.y, commit 5e8438fd7f11 ("scsi: replace blk_mq_pci_map_queues with blk_mq_map_hw_queues")
> > > was pulled in as depandency, the fix a9ae6fe1c319 ("blk-mq: create correct map for fallback case")
> > > should have just used 1452e9b470c9 ("blk-mq: introduce blk_mq_map_hw_queues")
> > > as Fixes, not the other conversion IMO.
> > 
> > What "other conversion"?  Sorry, I do not understand, did we take a
> > patch we shouldn't have, or did we miss a patch we should have applied?
> 
> If I understand the situation correctly, the problem is that v6.14.25
> ships commit 5e8438fd7f11 ("scsi: replace blk_mq_pci_map_queues with
> blk_mq_map_hw_queues") which introduced a regression for certain virtio
> configurations. The fixup patch is:
> 
> a9ae6fe1c319 ("blk-mq: create correct map for fallback case")
> 

But commit a9ae6fe1c319 ("blk-mq: create correct map for fallback case")
says it fixes commit a5665c3d150c ("virtio: blk/scsi: replace
blk_mq_virtio_map_queues with blk_mq_map_hw_queues"), which is ONLY in
the 6.14 tree.

Which is why we didn't pull it into the stable tree here.  Is that
commit just not marked properly?  Will it cause problems if it is
backported?

thanks,

greg k-h