Re: [PATCH] drm/xe: Add a per-client debugfs file reporting bo NUMA node distribution

["Thomas Hellström" <thomas.hellstrom@linux.intel.com>]

On Wed, 2026-10-28 at 16:07 +0100, Andi Shyti wrote:
> Hi Thomas,
> 
> On Wed, Oct 28, 2026 at 11:32:37AM +0100, Thomas Hellström wrote:
> > When debugging NUMA placement of bo system memory, there is
> > currently
> > no way to tell on which nodes the pages backing a bo reside. The
> > /proc/<pid>/numa_maps interface doesn't help since bo CPU mappings
> > are
> > VM_PFNMAP, and clients may have thousands of bos, so a dump of all
> > bos
> > is undesirable.
> 
> ...
> 
> > Example:
> >   exec 3<>/sys/kernel/debug/dri/client-<id>/bo_numa
> >   echo <handle1> >&3
> >   cat <&3
> >   echo <handle2> >&3
> >   cat <&3
> >   exec 3>&-
> > 
> > The client id can be obtained from the drm-client-id fdinfo key.
> > 
> > Cc: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
> > Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
> > Cc: Matthew Brost <matthew.brost@intel.com>
> > Cc: Matthew Auld <matthew.auld@intel.com>
> > Assisted-by: LLM
> 
> Please drop the Assisted-by and add your SoB :-)
> 
> ...
> 
> > +	do {
> > +		kvfree(buf);
> > +		bufsize *= 2;
> > +		buf = kvzalloc(bufsize, GFP_KERNEL_ACCOUNT);
> > +		if (!buf) {
> > +			ret = -ENOMEM;
> > +			goto out_unlock;
> > +		}
> > +
> > +		iter = (struct drm_print_iterator) {
> > +			.data = buf,
> > +			.remain = bufsize,
> > +		};
> > +		p = drm_coredump_printer(&iter);
> > +		ret = xe_bo_print_numa_info(file, handle, &p);
> > +		if (ret)
> > +			goto out_free;
> > +	} while (!iter.remain);
> 
> this is basically
> 
> 	} while(drm_coredump_printer_is_full(&p));
> 
> The rest looks good to me.
> 
> Andi

Thanks for reviewing, Andi. Will fix.

Thanks,
Thomas