From: Thierry Reding <treding@xxxxxxxxxx> The HOST1X_CHANNEL_DMAEND is an offset relative to the value written to the HOST1X_CHANNEL_DMASTART register, but it is currently treated as an absolute address. This can cause SMMU faults if the CDMA fetches past a pushbuffer's IOMMU mapping. Properly setting the DMAEND prevents the CDMA from fetching beyond that address and avoid such issues. This is currently not observed because a whole (almost) page of essentially scratch space absorbs any excessive prefetching by CDMA. However, changing the number of slots in the push buffer can trigger these SMMU faults. Signed-off-by: Thierry Reding <treding@xxxxxxxxxx> --- drivers/gpu/host1x/hw/cdma_hw.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/host1x/hw/cdma_hw.c b/drivers/gpu/host1x/hw/cdma_hw.c index 485aef5761af..a24c090ac96f 100644 --- a/drivers/gpu/host1x/hw/cdma_hw.c +++ b/drivers/gpu/host1x/hw/cdma_hw.c @@ -75,7 +75,7 @@ static void cdma_start(struct host1x_cdma *cdma) cdma->last_pos = cdma->push_buffer.pos; start = cdma->push_buffer.dma; - end = start + cdma->push_buffer.size + 4; + end = cdma->push_buffer.size + 4; host1x_ch_writel(ch, HOST1X_CHANNEL_DMACTRL_DMASTOP, HOST1X_CHANNEL_DMACTRL); @@ -126,7 +126,7 @@ static void cdma_timeout_restart(struct host1x_cdma *cdma, u32 getptr) HOST1X_CHANNEL_DMACTRL); start = cdma->push_buffer.dma; - end = start + cdma->push_buffer.size + 4; + end = cdma->push_buffer.size + 4; /* set base, end pointer (all of memory) */ host1x_ch_writel(ch, lower_32_bits(start), HOST1X_CHANNEL_DMASTART); -- 2.19.1 _______________________________________________ dri-devel mailing list dri-devel@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/dri-devel
