On Wed, 2025-02-26 at 17:47 +0000, Matthew Auld wrote:
Currently we just leave it uninitialised, which at first looks harmless, however we also don't zero out the pfn array, and with pfn_flags_mask the idea is to be able set individual flags for a given range of pfn or completely ignore them, outside of default_flags. So here we end up with pfn[i] & pfn_flags_mask, and if both are uninitialised we might get back an unexpected flags value, like asking for read only with default_flags, but getting back write on top, leading to potentially bogus behaviour.
To fix this ensure we zero the pfn_flags_mask, such that hmm only considers the default_flags and not also the initial pfn[i] value.
v2 (Thomas): - Prefer proper initializer.
Fixes: 81e058a3e7fd ("drm/xe: Introduce helper to populate userptr") Signed-off-by: Matthew Auld matthew.auld@intel.com Cc: Matthew Brost matthew.brost@intel.com Cc: Thomas Hellström thomas.hellstrom@intel.com Cc: stable@vger.kernel.org # v6.10+
drivers/gpu/drm/xe/xe_hmm.c | 18 ++++++++++-------- 1 file changed, 10 insertions(+), 8 deletions(-)
diff --git a/drivers/gpu/drm/xe/xe_hmm.c b/drivers/gpu/drm/xe/xe_hmm.c index 089834467880..2e4ae61567d8 100644 --- a/drivers/gpu/drm/xe/xe_hmm.c +++ b/drivers/gpu/drm/xe/xe_hmm.c @@ -166,13 +166,20 @@ int xe_hmm_userptr_populate_range(struct xe_userptr_vma *uvma, { unsigned long timeout = jiffies + msecs_to_jiffies(HMM_RANGE_DEFAULT_TIMEOUT);
- unsigned long *pfns, flags = HMM_PFN_REQ_FAULT;
- unsigned long *pfns;
struct xe_userptr *userptr; struct xe_vma *vma = &uvma->vma; u64 userptr_start = xe_vma_userptr(vma); u64 userptr_end = userptr_start + xe_vma_size(vma); struct xe_vm *vm = xe_vma_vm(vma);
- struct hmm_range hmm_range;
- struct hmm_range hmm_range = {
.pfn_flags_mask = 0, /* ignore pfns */.default_flags = HMM_PFN_REQ_FAULT,.start = userptr_start,.end = userptr_end,.notifier = &uvma->userptr.notifier,.dev_private_owner = vm->xe,- };
bool write = !xe_vma_read_only(vma); unsigned long notifier_seq; u64 npages; @@ -199,19 +206,14 @@ int xe_hmm_userptr_populate_range(struct xe_userptr_vma *uvma, return -ENOMEM; if (write)
flags |= HMM_PFN_REQ_WRITE;
hmm_range.default_flags |= HMM_PFN_REQ_WRITE;if (!mmget_not_zero(userptr->notifier.mm)) { ret = -EFAULT; goto free_pfns; }
- hmm_range.default_flags = flags;
hmm_range.hmm_pfns = pfns;
- hmm_range.notifier = &userptr->notifier;
- hmm_range.start = userptr_start;
- hmm_range.end = userptr_end;
- hmm_range.dev_private_owner = vm->xe;
while (true) { hmm_range.notifier_seq = mmu_interval_read_begin(&userptr->notifier);
Reviewed-by: Thomas Hellström thomas.hellstrom@linux.intel.com