diff mbox series

[v3] drm/i915: ensure segment offset never exceeds allowed max

Message ID luc7dpd4iag4nh4wvdhcfab34b2tcpryai3hgy22k3xmgvhyy7@xeglttypvlsf (mailing list archive)
State New
Headers show
Series [v3] drm/i915: ensure segment offset never exceeds allowed max | expand

Commit Message

Krzysztof Karas Oct. 28, 2024, 12:18 p.m. UTC 
Commit 255fc1703e42 ("drm/i915/gem: Calculate object page offset for
partial memory mapping") introduced a new offset that is compared to
sg_dma_len(r.sgt.sgp) in remap_io_sg() function. However, later in
remap_sg() the offset (which at that point resides in r->sgt.curr)
is compared to r->sgt.max. Scatter-gather list's max relies on one
of two values (see i915_scatterlist.h):
 a) sg_dma_len(s.sgp) when `dma` is true,
 b) s.sgp->length otherwise.
This suggests that in cases where `dma` is false, we should use
s.sgp->length to determine the max value instead of sg_dma_len(),
which is used regardless in remap_io_sg() (use_dma(iobase) might return
false there).

This patch uses r.sgt.max to check if offset is within allowed bounds,
because that max value is already set according to the `dma` setting.

v3:
 - instead of checking if r.sgt.curr would exceed allowed max, changed
the value in the while loop to be aligned with `dma` value.

Signed-off-by: Krzysztof Karas <krzysztof.karas@intel.com>
---
 drivers/gpu/drm/i915/i915_mm.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
diff mbox series

Patch

diff --git a/drivers/gpu/drm/i915/i915_mm.c b/drivers/gpu/drm/i915/i915_mm.c
index 8a2779191f18..3723291f2fef 100644
--- a/drivers/gpu/drm/i915/i915_mm.c
+++ b/drivers/gpu/drm/i915/i915_mm.c
@@ -143,8 +143,8 @@  int remap_io_sg(struct vm_area_struct *vma,
 	/* We rely on prevalidation of the io-mapping to skip track_pfn(). */
 	GEM_BUG_ON((vma->vm_flags & EXPECTED_FLAGS) != EXPECTED_FLAGS);
 
-	while (offset >= sg_dma_len(r.sgt.sgp) >> PAGE_SHIFT) {
-		offset -= sg_dma_len(r.sgt.sgp) >> PAGE_SHIFT;
+	while (offset >= r.sgt.max >> PAGE_SHIFT) {
+		offset -= r.sgt.max >> PAGE_SHIFT;
 		r.sgt = __sgt_iter(__sg_next(r.sgt.sgp), use_dma(iobase));
 		if (GEM_WARN_ON(!r.sgt.sgp))
 			return -EINVAL;