Re: [PATCH] drm/xe/vm: Avoid reserving zero fences

[Matthew Brost <matthew.brost@intel.com>]

On Thu, Feb 08, 2024 at 04:19:08PM +0100, Thomas Hellström wrote:
> On Thu, 2024-02-08 at 15:05 +0000, Matthew Auld wrote:
> > On 08/02/2024 13:21, Thomas Hellström wrote:
> > > The function xe_vm_prepare_vma was blindly accepting zero as the
> > > number of fences and forwarded that to drm_exec_prepare_obj.
> > > 
> > > However, that leads to an out-of-bounds shift in the
> > > dma_resv_reserve_fences() and while one could argue that the
> > > dma_resv code should be robust against that, avoid attempting
> > > to reserve zero fences.

I think the dma-resv should protect for warn against this. Seeems to be
a common mistake that more or less is silent. We've made this mistake at
least 2 times now [1] in Xe.

Matt

[1] https://patchwork.freedesktop.org/patch/572338/?series=127892&rev=2

> > > 
> > > Relevant stack trace:
> > > 
> > > [773.183188] ------------[ cut here ]------------
> > > [773.183199] UBSAN: shift-out-of-bounds in
> > > ../include/linux/log2.h:57:13
> > > [773.183241] shift exponent 64 is too large for 64-bit type 'long
> > > unsigned int'
> > > [773.183254] CPU: 2 PID: 1816 Comm: xe_evict Tainted: G    
> > > U             6.8.0-rc3-xe #1
> > > [773.183256] Hardware name: ASUS System Product Name/PRIME Z690-P
> > > D4, BIOS 2014 10/14/2022
> > > [773.183257] Call Trace:
> > > [773.183258]  <TASK>
> > > [773.183260]  dump_stack_lvl+0xaf/0xd0
> > > [773.183266]  dump_stack+0x10/0x20
> > > [773.183283]  ubsan_epilogue+0x9/0x40
> > > [773.183286]  __ubsan_handle_shift_out_of_bounds+0x10f/0x170
> > > [773.183293]  dma_resv_reserve_fences.cold+0x2b/0x48
> > > [773.183295]  ? ww_mutex_lock+0x3c/0x110
> > > [773.183301]  drm_exec_prepare_obj+0x45/0x60 [drm_exec]
> > > [773.183313]  xe_vm_prepare_vma+0x33/0x70 [xe]
> > > [773.183375]  xe_vma_destroy_unlocked+0x55/0xa0 [xe]
> > > [773.183427]  xe_vm_close_and_put+0x526/0x940 [xe]
> > > 
> > > Fixes: 2714d5093620 ("drm/xe: Convert pagefaulting code to use
> > > drm_exec")
> > > Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
> > > Cc: Matthew Brost <matthew.brost@intel.com>
> > > Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
> > > Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
> > Reviewed-by: Matthew Auld <matthew.auld@intel.com>
> 
> Thanks, Matthew
> What happened to your dma-resv patch you had for this?
> 
> 
> /Thomas