Ilia Mirkin
2014-Feb-16 04:27 UTC
[Nouveau] [PATCH] drm/nouveau: fix TTM_PL_TT memtype on pre-nv50
Commit a55409066 ("drm/nv50-: map TTM_PL_SYSTEM through a BAR for CPU access") made it possible to work with tiled memory. However mem->mm_node is not a nouveau_mem for AGP-using pre-NV50 cards, but a drm_mm_node, as created by the ttm_bo_manager_func. As such, extend the untiled check to explicitly include all pre-nv50 cards. Reported-by: Ronald <ronald645 at gmail.com> Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=74613 Signed-off-by: Ilia Mirkin <imirkin at alum.mit.edu> --- Hmmm... this seems like a really fragile semantic, I wonder if more mem->mm_node usages have to be audited. But this one's quick and easy. I'd rather see something a little more solid around dealing with these differences, but not sure how without reading all the ttm gunk. Perhaps just make sure it's always a nouveau_mem by wrapping the ttm_bo_manager logic? drivers/gpu/drm/nouveau/nouveau_bo.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/nouveau/nouveau_bo.c b/drivers/gpu/drm/nouveau/nouveau_bo.c index 488686d..4aed171 100644 --- a/drivers/gpu/drm/nouveau/nouveau_bo.c +++ b/drivers/gpu/drm/nouveau/nouveau_bo.c @@ -1249,7 +1249,7 @@ nouveau_ttm_io_mem_reserve(struct ttm_bo_device *bdev, struct ttm_mem_reg *mem) mem->bus.is_iomem = !dev->agp->cant_use_aperture; } #endif - if (!node->memtype) + if (nv_device(drm->device)->card_type < NV_50 || !node->memtype) /* untiled */ break; /* fallthrough, tiled memory */ -- 1.8.3.2
Ronald
2014-Feb-16 09:29 UTC
[Nouveau] [PATCH] drm/nouveau: fix TTM_PL_TT memtype on pre-nv50
2014-02-16 5:27 GMT+01:00 Ilia Mirkin <imirkin at alum.mit.edu>:> Commit a55409066 ("drm/nv50-: map TTM_PL_SYSTEM through a BAR for CPU > access") made it possible to work with tiled memory. However > mem->mm_node is not a nouveau_mem for AGP-using pre-NV50 cards, but a > drm_mm_node, as created by the ttm_bo_manager_func. As such, extend the > untiled check to explicitly include all pre-nv50 cards. > > Reported-by: Ronald <ronald645 at gmail.com> > Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=74613 > Signed-off-by: Ilia Mirkin <imirkin at alum.mit.edu> > ---Tested-by: Ronald Uitermark <ronald645 at gmail.com>> Hmmm... this seems like a really fragile semantic, I wonder if more > mem->mm_node usages have to be audited. But this one's quick and easy. I'd > rather see something a little more solid around dealing with these > differences, but not sure how without reading all the ttm gunk. Perhaps just > make sure it's always a nouveau_mem by wrapping the ttm_bo_manager logic? > > drivers/gpu/drm/nouveau/nouveau_bo.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/drivers/gpu/drm/nouveau/nouveau_bo.c b/drivers/gpu/drm/nouveau/nouveau_bo.c > index 488686d..4aed171 100644 > --- a/drivers/gpu/drm/nouveau/nouveau_bo.c > +++ b/drivers/gpu/drm/nouveau/nouveau_bo.c > @@ -1249,7 +1249,7 @@ nouveau_ttm_io_mem_reserve(struct ttm_bo_device *bdev, struct ttm_mem_reg *mem) > mem->bus.is_iomem = !dev->agp->cant_use_aperture; > } > #endif > - if (!node->memtype) > + if (nv_device(drm->device)->card_type < NV_50 || !node->memtype) > /* untiled */ > break; > /* fallthrough, tiled memory */ > -- > 1.8.3.2 >
Maarten Lankhorst
2014-Feb-16 10:43 UTC
[Nouveau] [PATCH] drm/nouveau: fix TTM_PL_TT memtype on pre-nv50
op 16-02-14 05:27, Ilia Mirkin schreef:> Commit a55409066 ("drm/nv50-: map TTM_PL_SYSTEM through a BAR for CPU > access") made it possible to work with tiled memory. However > mem->mm_node is not a nouveau_mem for AGP-using pre-NV50 cards, but a > drm_mm_node, as created by the ttm_bo_manager_func. As such, extend the > untiled check to explicitly include all pre-nv50 cards. > > Reported-by: Ronald <ronald645 at gmail.com> > Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=74613 > Signed-off-by: Ilia Mirkin <imirkin at alum.mit.edu> > --- > > Hmmm... this seems like a really fragile semantic, I wonder if more > mem->mm_node usages have to be audited. But this one's quick and easy. I'd > rather see something a little more solid around dealing with these > differences, but not sure how without reading all the ttm gunk. Perhaps just > make sure it's always a nouveau_mem by wrapping the ttm_bo_manager logic? > > drivers/gpu/drm/nouveau/nouveau_bo.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/drivers/gpu/drm/nouveau/nouveau_bo.c b/drivers/gpu/drm/nouveau/nouveau_bo.c > index 488686d..4aed171 100644 > --- a/drivers/gpu/drm/nouveau/nouveau_bo.c > +++ b/drivers/gpu/drm/nouveau/nouveau_bo.c > @@ -1249,7 +1249,7 @@ nouveau_ttm_io_mem_reserve(struct ttm_bo_device *bdev, struct ttm_mem_reg *mem) > mem->bus.is_iomem = !dev->agp->cant_use_aperture; > } > #endif > - if (!node->memtype) > + if (nv_device(drm->device)->card_type < NV_50 || !node->memtype) > /* untiled */ > break; > /* fallthrough, tiled memory */Acked-by: Maarten Lankhorst <maarten.lankhorst at canonical.com>
Apparently Analagous Threads
- [Bug 74613] New: [v3.14-rc1] [nv34] nouveau: get 0x10000000 put 0x0000ed30 state 0xc0000000 (err: MEM_FAULT) push 0x00000000
- [PATCH 1/7] drm/nouveau: fix m2mf copy to tiled gart
- [PATCH 8/8] drm/ttm: do not keep GPU dependent addresses
- [PATCH 2/7] drm/nv50-: untile mmap'd bo's
- [PATCH 8/8] drm/ttm: do not keep GPU dependent addresses