| Message ID | 20240517184811.25807-2-ian.forbes@broadcom.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | Fix memory limits for STDU | expand |
Hi Ian, kernel test robot noticed the following build warnings: [auto build test WARNING on drm/drm-next] [also build test WARNING on drm-exynos/exynos-drm-next drm-intel/for-linux-next drm-intel/for-linux-next-fixes drm-misc/drm-misc-next drm-tip/drm-tip linus/master v6.9 next-20240517] [If your patch is applied to the wrong git tree, kindly drop us a note. And when submitting patch, we suggest to use '--base' as documented in https://git-scm.com/docs/git-format-patch#_base_tree_information] url: https://github.com/intel-lab-lkp/linux/commits/Ian-Forbes/drm-vmwgfx-Filter-modes-which-exceed-graphics-memory/20240518-024955 base: git://anongit.freedesktop.org/drm/drm drm-next patch link: https://lore.kernel.org/r/20240517184811.25807-2-ian.forbes%40broadcom.com patch subject: [PATCH v2 1/4] drm/vmwgfx: Filter modes which exceed graphics memory config: i386-randconfig-141-20240518 (https://download.01.org/0day-ci/archive/20240518/202405181844.pA39DlHd-lkp@intel.com/config) compiler: clang version 18.1.5 (https://github.com/llvm/llvm-project 617a15a9eac96088ae5e9134248d8236e34b91b1) reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20240518/202405181844.pA39DlHd-lkp@intel.com/reproduce) If you fix the issue in a separate patch/commit (i.e. not just a new version of the same patch/commit), kindly add following tags | Reported-by: kernel test robot <lkp@intel.com> | Closes: https://lore.kernel.org/oe-kbuild-all/202405181844.pA39DlHd-lkp@intel.com/ All warnings (new ones prefixed by >>): >> drivers/gpu/drm/vmwgfx/vmwgfx_stdu.c:47: warning: This comment starts with '/**', but isn't a kernel-doc comment. Refer Documentation/doc-guide/kernel-doc.rst * llvmpipe will align the width and height of its buffers to match its vim +47 drivers/gpu/drm/vmwgfx/vmwgfx_stdu.c 38 39 #define vmw_crtc_to_stdu(x) \ 40 container_of(x, struct vmw_screen_target_display_unit, base.crtc) 41 #define vmw_encoder_to_stdu(x) \ 42 container_of(x, struct vmw_screen_target_display_unit, base.encoder) 43 #define vmw_connector_to_stdu(x) \ 44 container_of(x, struct vmw_screen_target_display_unit, base.connector) 45 46 /** > 47 * llvmpipe will align the width and height of its buffers to match its 48 * tile size. We need to keep this in mind when exposing modes to userspace 49 * so that this possible over-allocation will not exceed graphics memory. 50 */ 51 #define LLVM_PIPE_TILE_SIZE 64 52
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_stdu.c b/drivers/gpu/drm/vmwgfx/vmwgfx_stdu.c index 2041c4d48daa..a2b5527a249d 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_stdu.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_stdu.c @@ -43,7 +43,12 @@ #define vmw_connector_to_stdu(x) \ container_of(x, struct vmw_screen_target_display_unit, base.connector) - +/** + * llvmpipe will align the width and height of its buffers to match its + * tile size. We need to keep this in mind when exposing modes to userspace + * so that this possible over-allocation will not exceed graphics memory. + */ +#define LLVM_PIPE_TILE_SIZE 64 enum stdu_content_type { SAME_AS_DISPLAY = 0, @@ -829,7 +834,41 @@ static void vmw_stdu_connector_destroy(struct drm_connector *connector) vmw_stdu_destroy(vmw_connector_to_stdu(connector)); } +static enum drm_mode_status +vmw_stdu_connector_mode_valid(struct drm_connector *connector, + struct drm_display_mode *mode) +{ + enum drm_mode_status ret; + struct drm_device *dev = connector->dev; + struct vmw_private *dev_priv = vmw_priv(dev); + u64 assumed_cpp = dev_priv->assume_16bpp ? 2 : 4; + /* Align width and height to account for llvmpipe tile over-alignment */ + u64 required_mem = ALIGN(mode->hdisplay, LLVM_PIPE_TILE_SIZE) * + ALIGN(mode->vdisplay, LLVM_PIPE_TILE_SIZE) * + assumed_cpp; + required_mem = ALIGN(required_mem, PAGE_SIZE); + + ret = drm_mode_validate_size(mode, dev_priv->stdu_max_width, + dev_priv->stdu_max_height); + if (ret != MODE_OK) + return ret; + ret = drm_mode_validate_size(mode, dev_priv->texture_max_width, + dev_priv->texture_max_height); + if (ret != MODE_OK) + return ret; + + if (required_mem > dev_priv->max_primary_mem) + return MODE_MEM; + + if (required_mem > dev_priv->max_mob_pages * PAGE_SIZE) + return MODE_MEM; + + if (required_mem > dev_priv->max_mob_size) + return MODE_MEM; + + return MODE_OK; +} static const struct drm_connector_funcs vmw_stdu_connector_funcs = { .dpms = vmw_du_connector_dpms, @@ -845,7 +884,7 @@ static const struct drm_connector_funcs vmw_stdu_connector_funcs = { static const struct drm_connector_helper_funcs vmw_stdu_connector_helper_funcs = { .get_modes = vmw_connector_get_modes, - .mode_valid = vmw_connector_mode_valid + .mode_valid = vmw_stdu_connector_mode_valid };
SVGA requires individual surfaces to fit within graphics memory (max_mob_pages) which means that modes with a final buffer size that would exceed graphics memory must be pruned otherwise creation will fail. Additionally llvmpipe requires its buffer height and width to be a multiple of its tile size which is 64. As a result we have to anticipate that llvmpipe will round up the mode size passed to it by the compositor when it creates buffers and filter modes where this rounding exceeds graphics memory. This fixes an issue where VMs with low graphics memory (< 64MiB) configured with high resolution mode boot to a black screen because surface creation fails. Fixes: d947d1b71deb ("drm/vmwgfx: Add and connect connector helper function") Signed-off-by: Ian Forbes <ian.forbes@broadcom.com> --- drivers/gpu/drm/vmwgfx/vmwgfx_stdu.c | 43 ++++++++++++++++++++++++++-- 1 file changed, 41 insertions(+), 2 deletions(-)