search for: gf100_fb_init

...diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gf108.c b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gf108.c index 76678dd60f93f..c4c6f67af7ccc 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gf108.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gf108.c @@ -31,6 +31,7 @@ gf108_fb = { .init = gf100_fb_init, .init_page = gf100_fb_init_page, .intr = gf100_fb_intr, + .sysmem.flush_page_init = gf100_fb_sysmem_flush_page_init, .ram_new = gf108_ram_new, .default_bigpage = 17, }; diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gk104.c b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gk104.c index f7344...

After observing random crashes on Tegra devices when patch "fb/gf100: defer DMA mapping of scratch page to oneinit() hook" was applied, I noticed that moving the 100c10 page allocation to the oneinit() hook resulted in that page being now allocated for Tegra as well, and accesses be made to members of gf100_fb which were not accessible because gk20a_fb_new() called nvkm_fb_new_()

...nged, 1 insertion(+), 1 deletion(-) diff --git a/drm/nouveau/nvkm/subdev/fb/gm107.c b/drm/nouveau/nvkm/subdev/fb/gm107.c index 2a91df8..9cc7e61 100644 --- a/drm/nouveau/nvkm/subdev/fb/gm107.c +++ b/drm/nouveau/nvkm/subdev/fb/gm107.c @@ -29,7 +29,7 @@ gm107_fb = { .dtor = gf100_fb_dtor, .init = gf100_fb_init, .intr = gf100_fb_intr, - .ram_new = gm107_ram_new, + .ram_new = gk104_ram_new, .memtype_valid = gf100_fb_memtype_valid, }; -- 2.7.2

...tions(+), 8 deletions(-) diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gf100.c b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gf100.c index 76433cc66fff..5c8132873e60 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gf100.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gf100.c @@ -93,7 +93,18 @@ gf100_fb_init(struct nvkm_fb *base) struct gf100_fb *fb = gf100_fb(base); struct nvkm_device *device = fb->base.subdev.device; - if (fb->r100c10_page) + if (!fb->r100c10) { + dma_addr_t addr = dma_map_page(device->dev, fb->r100c10_page, 0, + PAGE_SIZE, DMA_BIDIRECTIONAL); + if (...

...32(device, 0x100ccc, nvkm_memory_addr(fb->mmu_rd) >> 8); -} - +/* GK20A's FB is similar to GF100's, but without the ability to allocate VRAM */ static const struct nvkm_fb_func gk20a_fb = { + .dtor = gf100_fb_dtor, .oneinit = gf100_fb_oneinit, - .init = gk20a_fb_init, + .init = gf100_fb_init, .init_page = gf100_fb_init_page, + .intr = gf100_fb_intr, .memtype_valid = gf100_fb_memtype_valid, }; -- 2.10.0

..._0 }, > + { gk104_fb_clkgate_blcg_init_main_0 }, > + { gk104_fb_clkgate_blcg_init_bcast_0 }, > + {} > +}; > + > +static const struct nvkm_fb_func > +gk110_fb = { > + .dtor = gf100_fb_dtor, > + .oneinit = gf100_fb_oneinit, > + .init = gf100_fb_init, > + .init_page = gf100_fb_init_page, > + .intr = gf100_fb_intr, > + .ram_new = gk104_ram_new, > + .default_bigpage = 17, > + .clkgate_pack = gk110_fb_clkgate_pack, > +}; > + > +int > +gk110_fb_new(struct nvkm_device *device, int index, struc...

This v4 is now a 3 piece series, after Alexandre pointed out that both GF 100 and NV50 are affected by the same issue, and that a related issue has been solved already for Tegra in commit 9d0394c6bed5 ("drm/nouveau/instmem/gk20a: set DMA mask early"). The issue that this series addresses is the fact that the Nouveau driver invokes the DMA API before setting the DMA mask. In both cases

A quick look over the memory reclocking part inside the maxwell mmiotraces kind of reminded me of the kepler memory reclocking, it looks for most parts the same. I didn't do a deep inspection so treat this with great care, but it seems to work at least on those maxwell gpus this was tested on. It depends on the PMU fixes for the fuc5 though, otherwise the PMU might not work as expected and

...nvkm-y += nvkm/subdev/fb/gp100.o nvkm-y += nvkm/subdev/fb/gp104.o diff --git a/drm/nouveau/nvkm/subdev/fb/gf100.h b/drm/nouveau/nvkm/subdev/fb/gf100.h index 449f431644b3..412eb89834e8 100644 --- a/drm/nouveau/nvkm/subdev/fb/gf100.h +++ b/drm/nouveau/nvkm/subdev/fb/gf100.h @@ -16,4 +16,8 @@ void gf100_fb_init(struct nvkm_fb *); void gf100_fb_intr(struct nvkm_fb *); void gp100_fb_init(struct nvkm_fb *); + +void gm200_fb_init_page(struct nvkm_fb *fb); +void gm200_fb_init(struct nvkm_fb *base); + #endif diff --git a/drm/nouveau/nvkm/subdev/fb/gm200.c b/drm/nouveau/nvkm/subdev/fb/gm200.c index 62f65324...

...e_pack[] = { + { gk110_fb_clkgate_blcg_init_unk_0 }, + { gk104_fb_clkgate_blcg_init_vm_0 }, + { gk104_fb_clkgate_blcg_init_main_0 }, + { gk104_fb_clkgate_blcg_init_bcast_0 }, + {} +}; + +static const struct nvkm_fb_func +gk110_fb = { + .dtor = gf100_fb_dtor, + .oneinit = gf100_fb_oneinit, + .init = gf100_fb_init, + .init_page = gf100_fb_init_page, + .intr = gf100_fb_intr, + .ram_new = gk104_ram_new, + .default_bigpage = 17, + .clkgate_pack = gk110_fb_clkgate_pack, +}; + +int +gk110_fb_new(struct nvkm_device *device, int index, struct nvkm_fb **pfb) +{ + return gf100_fb_new_(&gk110_fb, device, index, pf...

...2cc6b 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gf100.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gf100.c @@ -26,6 +26,7 @@ #include <core/memory.h> #include <core/option.h> +#include <subdev/therm.h> void gf100_fb_intr(struct nvkm_fb *base) @@ -92,6 +93,11 @@ gf100_fb_init(struct nvkm_fb *base) if (fb->r100c10_page) nvkm_wr32(device, 0x100c10, fb->r100c10 >> 8); + + if (base->func->clkgate_pack) { + nvkm_therm_clkgate_init(device->therm, + base->func->clkgate_pack); + } } void * diff --git a/drivers/gpu/drm/nouveau/nvkm/subd...

Next version of my patchseries for adding clockgating support for kepler1 and 2 on nouveau. The first version of this series can be found here: https://patchwork.freedesktop.org/series/36504/ Some minor changes: - Clarified that SLCG stands for 'secondary level clockgating', thanks for the small tip nvidia! - Removed the concept of levels, this was more useful for debugging then

Next version of my patchseries for adding clockgating support for kepler1 and 2 on nouveau. The first version of this series can be found here: https://patchwork.freedesktop.org/series/36504/ One small change: - Set therm->clkgate_enabled to false until the last patch, where we introduce the NvPmEnableGating option Lyude Paul (5): drm/nouveau: Add support for basic clockgating on

Next version of my patchseries for adding clockgating support for kepler1 and 2 on nouveau. The first version of this series can be found here: https://patchwork.freedesktop.org/series/36504/ One small change: - Add Martin's R-B, whoops Lyude Paul (5): drm/nouveau: Add support for basic clockgating on Kepler1 drm/nouveau: Add support for BLCG on Kepler1 drm/nouveau: Add support for

Next version of my patchseries for adding clockgating support for kepler1 and 2 on nouveau. The first version of this series can be found here: https://patchwork.freedesktop.org/series/36504/ Some small changes (shouldn't change anything functionally): - Made gf100_therm_new, gk104_clkgate_pack, and gk110_fb_clkgate_blcg_init_unk_0 static (re: kbuild test bot) - Making all three of

Next version of my patchseries for adding clockgating support for kepler1 and 2 on nouveau. The first version of this series can be found here: https://patchwork.freedesktop.org/series/36504/ Some very important changes: - Fix gf100_clkgate_init() to actually write registers! This got broken in the last version by accident - Dump the register packs and the resulting register writes into

It's here! After a lot of investigation, rewrites, and traces, I present the patch series to implement all known levels of clockgating for Kepler1 and Kepler2 GPUs. Starting with Fermi GPUs (this is probably present on earlier GPUs as well, but with a far less easy to manage interface), nvidia added two clockgating levels that are handled mostly in firmware (with the exception of course, of