From eec7e23d848d2194dd8791fcd0f4a54d4378eecd Mon Sep 17 00:00:00 2001 From: Akash Goel Date: Thu, 27 Nov 2025 16:49:12 +0000 Subject: drm/panthor: Prevent potential UAF in group creation This commit prevents the possibility of a use after free issue in the GROUP_CREATE ioctl function, which arose as pointer to the group is accessed in that ioctl function after storing it in the Xarray. A malicious userspace can second guess the handle of a group and try to call GROUP_DESTROY ioctl from another thread around the same time as GROUP_CREATE ioctl. To prevent the use after free exploit, this commit uses a mark on an entry of group pool Xarray which is added just before returning from the GROUP_CREATE ioctl function. The mark is checked for all ioctls that specify the group handle and so userspace won't be abe to delete a group that isn't marked yet. v2: Add R-bs and fixes tags Fixes: de85488138247 ("drm/panthor: Add the scheduler logical block") Co-developed-by: Boris Brezillon Signed-off-by: Boris Brezillon Signed-off-by: Akash Goel Reviewed-by: Boris Brezillon Reviewed-by: Steven Price Reviewed-by: Chia-I Wu Link: https://patch.msgid.link/20251127164912.3788155-1-akash.goel@arm.com --- drivers/gpu/drm/panthor/panthor_sched.c | 19 +++++++++++++++---- 1 file changed, 15 insertions(+), 4 deletions(-) diff --git a/drivers/gpu/drm/panthor/panthor_sched.c b/drivers/gpu/drm/panthor/panthor_sched.c index e74ca071159d..5d2e9144d450 100644 --- a/drivers/gpu/drm/panthor/panthor_sched.c +++ b/drivers/gpu/drm/panthor/panthor_sched.c @@ -776,6 +776,12 @@ struct panthor_job_profiling_data { */ #define MAX_GROUPS_PER_POOL 128 +/* + * Mark added on an entry of group pool Xarray to identify if the group has + * been fully initialized and can be accessed elsewhere in the driver code. + */ +#define GROUP_REGISTERED XA_MARK_1 + /** * struct panthor_group_pool - Group pool * @@ -2906,7 +2912,7 @@ void panthor_fdinfo_gather_group_samples(struct panthor_file *pfile) return; xa_lock(&gpool->xa); - xa_for_each(&gpool->xa, i, group) { + xa_for_each_marked(&gpool->xa, i, group, GROUP_REGISTERED) { guard(spinlock)(&group->fdinfo.lock); pfile->stats.cycles += group->fdinfo.data.cycles; pfile->stats.time += group->fdinfo.data.time; @@ -3591,6 +3597,8 @@ int panthor_group_create(struct panthor_file *pfile, group_init_task_info(group); + xa_set_mark(&gpool->xa, gid, GROUP_REGISTERED); + return gid; err_erase_gid: @@ -3608,6 +3616,9 @@ int panthor_group_destroy(struct panthor_file *pfile, u32 group_handle) struct panthor_scheduler *sched = ptdev->scheduler; struct panthor_group *group; + if (!xa_get_mark(&gpool->xa, group_handle, GROUP_REGISTERED)) + return -EINVAL; + group = xa_erase(&gpool->xa, group_handle); if (!group) return -EINVAL; @@ -3633,12 +3644,12 @@ int panthor_group_destroy(struct panthor_file *pfile, u32 group_handle) } static struct panthor_group *group_from_handle(struct panthor_group_pool *pool, - u32 group_handle) + unsigned long group_handle) { struct panthor_group *group; xa_lock(&pool->xa); - group = group_get(xa_load(&pool->xa, group_handle)); + group = group_get(xa_find(&pool->xa, &group_handle, group_handle, GROUP_REGISTERED)); xa_unlock(&pool->xa); return group; @@ -3725,7 +3736,7 @@ panthor_fdinfo_gather_group_mem_info(struct panthor_file *pfile, return; xa_lock(&gpool->xa); - xa_for_each(&gpool->xa, i, group) { + xa_for_each_marked(&gpool->xa, i, group, GROUP_REGISTERED) { stats->resident += group->fdinfo.kbo_sizes; if (group->csg_id >= 0) stats->active += group->fdinfo.kbo_sizes; -- cgit From 56715b45e3bfecf82e6404df19a10699a67c3a74 Mon Sep 17 00:00:00 2001 From: Ludovic Desroches Date: Wed, 26 Nov 2025 15:44:44 +0100 Subject: drm/gem-dma: revert the 8-byte alignment constraint MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Using drm_mode_size_dumb() to compute the size of dumb buffers introduced an 8-byte alignment constraint on the pitch that wasn’t present before. Let’s remove this constraint, which isn’t necessarily required and may cause buffers to be allocated larger than needed. Signed-off-by: Ludovic Desroches Reviewed-by: Thomas Zimmermann Fixes: dcacfcd35cef ("drm/gem-dma: Compute dumb-buffer sizes with drm_mode_size_dumb()") Signed-off-by: Thomas Zimmermann Link: https://patch.msgid.link/20251126-lcd_pitch_alignment-v1-1-991610a1e369@microchip.com --- drivers/gpu/drm/drm_gem_dma_helper.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/drm_gem_dma_helper.c b/drivers/gpu/drm/drm_gem_dma_helper.c index 12d8307997a0..eb56ba234796 100644 --- a/drivers/gpu/drm/drm_gem_dma_helper.c +++ b/drivers/gpu/drm/drm_gem_dma_helper.c @@ -308,7 +308,7 @@ int drm_gem_dma_dumb_create(struct drm_file *file_priv, struct drm_gem_dma_object *dma_obj; int ret; - ret = drm_mode_size_dumb(drm, args, SZ_8, 0); + ret = drm_mode_size_dumb(drm, args, 0, 0); if (ret) return ret; -- cgit From 4cbae3748fc479f1944e8ef2b3ed4a05cd799ce8 Mon Sep 17 00:00:00 2001 From: Ludovic Desroches Date: Wed, 26 Nov 2025 15:44:45 +0100 Subject: drm/gem-shmem: revert the 8-byte alignment constraint MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Using drm_mode_size_dumb() to compute the size of dumb buffers introduced an 8-byte alignment constraint on the pitch that wasn’t present before. Let’s remove this constraint, which isn’t necessarily required and may cause buffers to be allocated larger than needed. Signed-off-by: Ludovic Desroches Reviewed-by: Thomas Zimmermann Fixes: 4977dcecb931 ("drm/gem-shmem: Compute dumb-buffer sizes with drm_mode_size_dumb()") Signed-off-by: Thomas Zimmermann Link: https://patch.msgid.link/20251126-lcd_pitch_alignment-v1-2-991610a1e369@microchip.com --- drivers/gpu/drm/drm_gem_shmem_helper.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/drm_gem_shmem_helper.c b/drivers/gpu/drm/drm_gem_shmem_helper.c index dc94a27710e5..93b9cff89080 100644 --- a/drivers/gpu/drm/drm_gem_shmem_helper.c +++ b/drivers/gpu/drm/drm_gem_shmem_helper.c @@ -559,7 +559,7 @@ int drm_gem_shmem_dumb_create(struct drm_file *file, struct drm_device *dev, { int ret; - ret = drm_mode_size_dumb(dev, args, SZ_8, 0); + ret = drm_mode_size_dumb(dev, args, 0, 0); if (ret) return ret; -- cgit From 308eeb8ca3fdee54bcd3ec7272c05688e178299a Mon Sep 17 00:00:00 2001 From: Marek Vasut Date: Tue, 2 Dec 2025 19:11:06 +0100 Subject: drm/rcar-du: dsi: Handle both DRM_MODE_FLAG_N.SYNC and !DRM_MODE_FLAG_P.SYNC Since commit 94fe479fae96 ("drm/rcar-du: dsi: Clean up handling of DRM mode flags") the driver does not set TXVMVPRMSET0R_VSPOL_LOW and TXVMVPRMSET0R_HSPOL_LOW for modes which set neither DRM_MODE_FLAG_[PN].SYNC. The previous behavior was to assume that neither flag means DRM_MODE_FLAG_N.SYNC . Restore the previous behavior for maximum compatibility. The change of behavior is visible below, consider Vertical mode->flags for simplicity sake, although the same applies to Horizontal ones: Before 94fe479fae96 ("drm/rcar-du: dsi: Clean up handling of DRM mode flags") : - DRM_MODE_FLAG_PVSYNC => vprmset0r |= 0 - DRM_MODE_FLAG_NVSYNC => vprmset0r |= TXVMVPRMSET0R_VSPOL_LOW - Neither DRM_MODE_FLAG_[PN]VSYNC => vprmset0r |= TXVMVPRMSET0R_VSPOL_LOW After 94fe479fae96 ("drm/rcar-du: dsi: Clean up handling of DRM mode flags") : - DRM_MODE_FLAG_PVSYNC => vprmset0r |= 0 - DRM_MODE_FLAG_NVSYNC => vprmset0r |= TXVMVPRMSET0R_VSPOL_LOW - Neither DRM_MODE_FLAG_[PN]VSYNC => vprmset0r |= 0 <---------- This broke The "Neither" case behavior is different, because DRM_MODE_FLAG_N[HV]SYNC is really not equivalent !DRM_MODE_FLAG_P[HV]SYNC . Fixes: 94fe479fae96 ("drm/rcar-du: dsi: Clean up handling of DRM mode flags") Signed-off-by: Marek Vasut Reviewed-by: Laurent Pinchart Link: https://patch.msgid.link/20251202181146.138365-1-marek.vasut+renesas@mailbox.org Signed-off-by: Tomi Valkeinen --- drivers/gpu/drm/renesas/rcar-du/rcar_mipi_dsi.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/renesas/rcar-du/rcar_mipi_dsi.c b/drivers/gpu/drm/renesas/rcar-du/rcar_mipi_dsi.c index 9413b76d0bfc..4ef2e3c129ed 100644 --- a/drivers/gpu/drm/renesas/rcar-du/rcar_mipi_dsi.c +++ b/drivers/gpu/drm/renesas/rcar-du/rcar_mipi_dsi.c @@ -492,9 +492,9 @@ static void rcar_mipi_dsi_set_display_timing(struct rcar_mipi_dsi *dsi, /* Configuration for Video Parameters, input is always RGB888 */ vprmset0r = TXVMVPRMSET0R_BPP_24; - if (mode->flags & DRM_MODE_FLAG_NVSYNC) + if (!(mode->flags & DRM_MODE_FLAG_PVSYNC)) vprmset0r |= TXVMVPRMSET0R_VSPOL_LOW; - if (mode->flags & DRM_MODE_FLAG_NHSYNC) + if (!(mode->flags & DRM_MODE_FLAG_PHSYNC)) vprmset0r |= TXVMVPRMSET0R_HSPOL_LOW; vprmset1r = TXVMVPRMSET1R_VACTIVE(mode->vdisplay) -- cgit From bf2084a7b1d75d093b6a79df4c10142d49fbaa0e Mon Sep 17 00:00:00 2001 From: Xiaogang Chen Date: Mon, 1 Dec 2025 14:12:29 -0600 Subject: drm/amdkfd: Use huge page size to check split svm range alignment When split svm ranges that have been mapped using huge page should use huge page size(2MB) to check split range alignment, not prange->granularity that means migration granularity. Fixes: 7ef6b2d4b7e5 ("drm/amdkfd: remap unaligned svm ranges that have split") Signed-off-by: Xiaogang Chen Reviewed-by: Philip Yang Signed-off-by: Alex Deucher (cherry picked from commit 448ee45353ef9fb1a34f5f26eb3f48923c6f0898) --- drivers/gpu/drm/amd/amdkfd/kfd_svm.c | 46 +++++++++++++++++++++++++----------- 1 file changed, 32 insertions(+), 14 deletions(-) diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c index 97c2270f278f..79ea138897fc 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c @@ -1144,30 +1144,48 @@ static int svm_range_split_tail(struct svm_range *prange, uint64_t new_last, struct list_head *insert_list, struct list_head *remap_list) { + unsigned long last_align_down = ALIGN_DOWN(prange->last, 512); + unsigned long start_align = ALIGN(prange->start, 512); + bool huge_page_mapping = last_align_down > start_align; struct svm_range *tail = NULL; - int r = svm_range_split(prange, prange->start, new_last, &tail); + int r; - if (!r) { - list_add(&tail->list, insert_list); - if (!IS_ALIGNED(new_last + 1, 1UL << prange->granularity)) - list_add(&tail->update_list, remap_list); - } - return r; + r = svm_range_split(prange, prange->start, new_last, &tail); + + if (r) + return r; + + list_add(&tail->list, insert_list); + + if (huge_page_mapping && tail->start > start_align && + tail->start < last_align_down && (!IS_ALIGNED(tail->start, 512))) + list_add(&tail->update_list, remap_list); + + return 0; } static int svm_range_split_head(struct svm_range *prange, uint64_t new_start, struct list_head *insert_list, struct list_head *remap_list) { + unsigned long last_align_down = ALIGN_DOWN(prange->last, 512); + unsigned long start_align = ALIGN(prange->start, 512); + bool huge_page_mapping = last_align_down > start_align; struct svm_range *head = NULL; - int r = svm_range_split(prange, new_start, prange->last, &head); + int r; - if (!r) { - list_add(&head->list, insert_list); - if (!IS_ALIGNED(new_start, 1UL << prange->granularity)) - list_add(&head->update_list, remap_list); - } - return r; + r = svm_range_split(prange, new_start, prange->last, &head); + + if (r) + return r; + + list_add(&head->list, insert_list); + + if (huge_page_mapping && head->last + 1 > start_align && + head->last + 1 < last_align_down && (!IS_ALIGNED(head->last, 512))) + list_add(&head->update_list, remap_list); + + return 0; } static void -- cgit From b7851f8c66191cd23a0a08bd484465ad74bbbb7d Mon Sep 17 00:00:00 2001 From: Jay Cornwall Date: Fri, 14 Nov 2025 14:32:42 -0600 Subject: drm/amdkfd: Trap handler support for expert scheduling mode The trap may be entered with dependency checking disabled. Wait for dependency counters and save/restore scheduling mode. v2: Use ttmp1 instead of ttmp11. ttmp11 is not zero-initialized. While the trap handler does zero this field before use, a user-mode second-level trap handler could not rely on this being zero when using an older kernel mode driver. v3: Use ttmp11 primarily but copy to ttmp1 before jumping to the second level trap handler. ttmp1 is inspectable by a debugger. Unexpected bits in the unused space may regress existing software. Signed-off-by: Jay Cornwall Reviewed-by: Lancelot Six Signed-off-by: Alex Deucher (cherry picked from commit 423888879412e94725ca2bdccd89414887d98e31) Cc: stable@vger.kernel.org --- drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler.h | 62 +++++++++++++--------- .../gpu/drm/amd/amdkfd/cwsr_trap_handler_gfx12.asm | 37 +++++++++++++ 2 files changed, 73 insertions(+), 26 deletions(-) diff --git a/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler.h b/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler.h index 0320163b6e74..f98c735b2905 100644 --- a/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler.h +++ b/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler.h @@ -3644,14 +3644,18 @@ static const uint32_t cwsr_trap_gfx9_4_3_hex[] = { }; static const uint32_t cwsr_trap_gfx12_hex[] = { - 0xbfa00001, 0xbfa002a2, - 0xb0804009, 0xb8f8f804, + 0xbfa00001, 0xbfa002b2, + 0xb0804009, 0xb8eef81a, + 0xbf880000, 0xb980081a, + 0x00000000, 0xb8f8f804, + 0x9177ff77, 0x0c000000, + 0x846e9a6e, 0x8c776e77, 0x9178ff78, 0x00008c00, 0xb8fbf811, 0x8b6eff78, 0x00004000, 0xbfa10008, 0x8b6eff7b, 0x00000080, 0xbfa20018, 0x8b6ea07b, - 0xbfa20042, 0xbf830010, + 0xbfa2004a, 0xbf830010, 0xb8fbf811, 0xbfa0fffb, 0x8b6eff7b, 0x00000bd0, 0xbfa20010, 0xb8eef812, @@ -3662,28 +3666,32 @@ static const uint32_t cwsr_trap_gfx12_hex[] = { 0xf0000000, 0xbfa20005, 0x8b6fff6f, 0x00000200, 0xbfa20002, 0x8b6ea07b, - 0xbfa2002c, 0xbefa4d82, + 0xbfa20034, 0xbefa4d82, 0xbf8a0000, 0x84fa887a, 0xbf0d8f7b, 0xbfa10002, 0x8c7bff7b, 0xffff0000, - 0xf4601bbd, 0xf8000010, - 0xbf8a0000, 0x846e976e, - 0x9177ff77, 0x00800000, - 0x8c776e77, 0xf4603bbd, - 0xf8000000, 0xbf8a0000, - 0xf4603ebd, 0xf8000008, - 0xbf8a0000, 0x8bee6e6e, - 0xbfa10001, 0xbe80486e, - 0x8b6eff6d, 0xf0000000, - 0xbfa20009, 0xb8eef811, - 0x8b6eff6e, 0x00000080, - 0xbfa20007, 0x8c78ff78, - 0x00004000, 0x80ec886c, - 0x82ed806d, 0xbfa00002, - 0x806c846c, 0x826d806d, - 0x8b6dff6d, 0x0000ffff, - 0x8bfe7e7e, 0x8bea6a6a, - 0x85788978, 0xb9783244, + 0x8b6eff77, 0x0c000000, + 0x916dff6d, 0x0c000000, + 0x8c6d6e6d, 0xf4601bbd, + 0xf8000010, 0xbf8a0000, + 0x846e976e, 0x9177ff77, + 0x00800000, 0x8c776e77, + 0xf4603bbd, 0xf8000000, + 0xbf8a0000, 0xf4603ebd, + 0xf8000008, 0xbf8a0000, + 0x8bee6e6e, 0xbfa10001, + 0xbe80486e, 0x8b6eff6d, + 0xf0000000, 0xbfa20009, + 0xb8eef811, 0x8b6eff6e, + 0x00000080, 0xbfa20007, + 0x8c78ff78, 0x00004000, + 0x80ec886c, 0x82ed806d, + 0xbfa00002, 0x806c846c, + 0x826d806d, 0x8b6dff6d, + 0x0000ffff, 0x8bfe7e7e, + 0x8bea6a6a, 0x85788978, + 0x936eff77, 0x0002001a, + 0xb96ef81a, 0xb9783244, 0xbe804a6c, 0xb8faf802, 0xbf0d987a, 0xbfa10001, 0xbfb00000, 0x8b6dff6d, @@ -3981,7 +3989,7 @@ static const uint32_t cwsr_trap_gfx12_hex[] = { 0x008ce800, 0x00000000, 0x807d817d, 0x8070ff70, 0x00000080, 0xbf0a7b7d, - 0xbfa2fff7, 0xbfa0016e, + 0xbfa2fff7, 0xbfa00171, 0xbef4007e, 0x8b75ff7f, 0x0000ffff, 0x8c75ff75, 0x00040000, 0xbef60080, @@ -4163,12 +4171,14 @@ static const uint32_t cwsr_trap_gfx12_hex[] = { 0xf8000074, 0xbf8a0000, 0x8b6dff6d, 0x0000ffff, 0x8bfe7e7e, 0x8bea6a6a, - 0xb97af804, 0xbe804ec2, - 0xbf94fffe, 0xbe804a6c, + 0x936eff77, 0x0002001a, + 0xb96ef81a, 0xb97af804, 0xbe804ec2, 0xbf94fffe, - 0xbfb10000, 0xbf9f0000, + 0xbe804a6c, 0xbe804ec2, + 0xbf94fffe, 0xbfb10000, 0xbf9f0000, 0xbf9f0000, 0xbf9f0000, 0xbf9f0000, + 0xbf9f0000, 0x00000000, }; static const uint32_t cwsr_trap_gfx9_5_0_hex[] = { diff --git a/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler_gfx12.asm b/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler_gfx12.asm index 5a1a1b1f897f..07999b4649de 100644 --- a/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler_gfx12.asm +++ b/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler_gfx12.asm @@ -78,9 +78,16 @@ var SQ_WAVE_EXCP_FLAG_PRIV_RESTORE_PART_2_SHIFT = SQ_WAVE_EXCP_FLAG_PRIV_ILLEGAL var SQ_WAVE_EXCP_FLAG_PRIV_RESTORE_PART_2_SIZE = SQ_WAVE_EXCP_FLAG_PRIV_HOST_TRAP_SHIFT - SQ_WAVE_EXCP_FLAG_PRIV_ILLEGAL_INST_SHIFT var SQ_WAVE_EXCP_FLAG_PRIV_RESTORE_PART_3_SHIFT = SQ_WAVE_EXCP_FLAG_PRIV_WAVE_START_SHIFT var SQ_WAVE_EXCP_FLAG_PRIV_RESTORE_PART_3_SIZE = 32 - SQ_WAVE_EXCP_FLAG_PRIV_RESTORE_PART_3_SHIFT + +var SQ_WAVE_SCHED_MODE_DEP_MODE_SHIFT = 0 +var SQ_WAVE_SCHED_MODE_DEP_MODE_SIZE = 2 + var BARRIER_STATE_SIGNAL_OFFSET = 16 var BARRIER_STATE_VALID_OFFSET = 0 +var TTMP11_SCHED_MODE_SHIFT = 26 +var TTMP11_SCHED_MODE_SIZE = 2 +var TTMP11_SCHED_MODE_MASK = 0xC000000 var TTMP11_DEBUG_TRAP_ENABLED_SHIFT = 23 var TTMP11_DEBUG_TRAP_ENABLED_MASK = 0x800000 @@ -160,8 +167,19 @@ L_JUMP_TO_RESTORE: s_branch L_RESTORE L_SKIP_RESTORE: + // Assume most relaxed scheduling mode is set. Save and revert to normal mode. + s_getreg_b32 ttmp2, hwreg(HW_REG_WAVE_SCHED_MODE) + s_wait_alu 0 + s_setreg_imm32_b32 hwreg(HW_REG_WAVE_SCHED_MODE, \ + SQ_WAVE_SCHED_MODE_DEP_MODE_SHIFT, SQ_WAVE_SCHED_MODE_DEP_MODE_SIZE), 0 + s_getreg_b32 s_save_state_priv, hwreg(HW_REG_WAVE_STATE_PRIV) //save STATUS since we will change SCC + // Save SCHED_MODE[1:0] into ttmp11[27:26]. + s_andn2_b32 ttmp11, ttmp11, TTMP11_SCHED_MODE_MASK + s_lshl_b32 ttmp2, ttmp2, TTMP11_SCHED_MODE_SHIFT + s_or_b32 ttmp11, ttmp11, ttmp2 + // Clear SPI_PRIO: do not save with elevated priority. // Clear ECC_ERR: prevents SQC store and triggers FATAL_HALT if setreg'd. s_andn2_b32 s_save_state_priv, s_save_state_priv, SQ_WAVE_STATE_PRIV_ALWAYS_CLEAR_MASK @@ -238,6 +256,13 @@ L_FETCH_2ND_TRAP: s_cbranch_scc0 L_NO_SIGN_EXTEND_TMA s_or_b32 ttmp15, ttmp15, 0xFFFF0000 L_NO_SIGN_EXTEND_TMA: +#if ASIC_FAMILY == CHIP_GFX12 + // Move SCHED_MODE[1:0] from ttmp11 to unused bits in ttmp1[27:26] (return PC_HI). + // The second-level trap will restore from ttmp1 for backwards compatibility. + s_and_b32 ttmp2, ttmp11, TTMP11_SCHED_MODE_MASK + s_andn2_b32 ttmp1, ttmp1, TTMP11_SCHED_MODE_MASK + s_or_b32 ttmp1, ttmp1, ttmp2 +#endif s_load_dword ttmp2, [ttmp14, ttmp15], 0x10 scope:SCOPE_SYS // debug trap enabled flag s_wait_idle @@ -287,6 +312,10 @@ L_EXIT_TRAP: // STATE_PRIV.BARRIER_COMPLETE may have changed since we read it. // Only restore fields which the trap handler changes. s_lshr_b32 s_save_state_priv, s_save_state_priv, SQ_WAVE_STATE_PRIV_SCC_SHIFT + + // Assume relaxed scheduling mode after this point. + restore_sched_mode(ttmp2) + s_setreg_b32 hwreg(HW_REG_WAVE_STATE_PRIV, SQ_WAVE_STATE_PRIV_SCC_SHIFT, \ SQ_WAVE_STATE_PRIV_POISON_ERR_SHIFT - SQ_WAVE_STATE_PRIV_SCC_SHIFT + 1), s_save_state_priv @@ -1043,6 +1072,9 @@ L_SKIP_BARRIER_RESTORE: s_and_b64 exec, exec, exec // Restore STATUS.EXECZ, not writable by s_setreg_b32 s_and_b64 vcc, vcc, vcc // Restore STATUS.VCCZ, not writable by s_setreg_b32 + // Assume relaxed scheduling mode after this point. + restore_sched_mode(s_restore_tmp) + s_setreg_b32 hwreg(HW_REG_WAVE_STATE_PRIV), s_restore_state_priv // SCC is included, which is changed by previous salu // Make barrier and LDS state visible to all waves in the group. @@ -1134,3 +1166,8 @@ function valu_sgpr_hazard end #endif end + +function restore_sched_mode(s_tmp) + s_bfe_u32 s_tmp, ttmp11, (TTMP11_SCHED_MODE_SHIFT | (TTMP11_SCHED_MODE_SIZE << 0x10)) + s_setreg_b32 hwreg(HW_REG_WAVE_SCHED_MODE), s_tmp +end -- cgit From 3c41114dcdabb7b25f5bc33273c6db9c7af7f4a7 Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Tue, 11 Nov 2025 11:17:22 -0500 Subject: drm/amd/display: Use GFP_ATOMIC in dc_create_plane_state() This can get called from an atomic context. Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/4470 Reviewed-by: Harry Wentland Signed-off-by: Alex Deucher (cherry picked from commit 8acdad9344cc7b4e7bc01f0dfea80093eb3768db) Cc: stable@vger.kernel.org --- drivers/gpu/drm/amd/display/dc/core/dc_surface.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_surface.c b/drivers/gpu/drm/amd/display/dc/core/dc_surface.c index 922f23557f5d..0971dfa25845 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_surface.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_surface.c @@ -86,7 +86,7 @@ uint8_t dc_plane_get_pipe_mask(struct dc_state *dc_state, const struct dc_plane struct dc_plane_state *dc_create_plane_state(const struct dc *dc) { struct dc_plane_state *plane_state = kvzalloc(sizeof(*plane_state), - GFP_KERNEL); + GFP_ATOMIC); if (NULL == plane_state) return NULL; -- cgit From eb296c09805ee37dd4ea520a7fb3ec157c31090f Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Tue, 2 Dec 2025 14:24:03 -0500 Subject: drm/amdgpu: don't attach the tlb fence for SI MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit SI hardware doesn't support pasids, user mode queues, or KIQ/MES so there is no need for this. Doing so results in a segfault as these callbacks are non-existent for SI. Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/4744 Fixes: f3854e04b708 ("drm/amdgpu: attach tlb fence to the PTs update") Reviewed-by: Timur Kristóf Signed-off-by: Alex Deucher (cherry picked from commit 820b3d376e8a102c6aeab737ec6edebbbb710e04) --- drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c index a67285118c37..c362d4dfb5bb 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c @@ -1069,7 +1069,9 @@ amdgpu_vm_tlb_flush(struct amdgpu_vm_update_params *params, } /* Prepare a TLB flush fence to be attached to PTs */ - if (!params->unlocked) { + if (!params->unlocked && + /* SI doesn't support pasid or KIQ/MES */ + params->adev->family > AMDGPU_FAMILY_SI) { amdgpu_vm_tlb_fence_create(params->adev, vm, fence); /* Makes sure no PD/PT is freed before the flush */ -- cgit From cf841f6abc26e455c68725e071450d6a29a53611 Mon Sep 17 00:00:00 2001 From: Alex Hung Date: Wed, 26 Nov 2025 11:44:35 -0700 Subject: drm/amd/display: Refactor dml_core_mode_support to reduce stack frame [WHAT] When compiling Linux kernel with clang, the following warning / error messages pops up: drivers/gpu/drm/amd/amdgpu/../dal-dev/dc/dml2_0/display_mode_core.c:6853:12: error: stack frame size (2120) exceeds limit (2056) in 'dml_core_mode_support' [-Werror,-Wframe-larger-than] 6853 | dml_bool_t dml_core_mode_support(struct display_mode_lib_st *mode_lib) [HOW] Refactoring CalculateVMRowAndSwath_params assignments to a new function helps reduce the stack frame size in dml_core_mode_support. Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/4733 Reviewed-by: Austin Zheng Reviewed-by: Aurabindo Pillai Signed-off-by: Alex Hung Signed-off-by: Chenyu Chen Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher (cherry picked from commit 29a4dc4b5d82e6b3da343391f9e784cf5c48732c) --- .../drm/amd/display/dc/dml2_0/display_mode_core.c | 134 +++++++++++---------- 1 file changed, 71 insertions(+), 63 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/display_mode_core.c b/drivers/gpu/drm/amd/display/dc/dml2_0/display_mode_core.c index c468f492b876..09303c282495 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2_0/display_mode_core.c +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/display_mode_core.c @@ -6711,6 +6711,76 @@ static noinline_for_stack void dml_prefetch_check(struct display_mode_lib_st *mo } // for j } +static noinline_for_stack void set_vm_row_and_swath_parameters(struct display_mode_lib_st *mode_lib) +{ + struct CalculateVMRowAndSwath_params_st *CalculateVMRowAndSwath_params = &mode_lib->scratch.CalculateVMRowAndSwath_params; + struct dml_core_mode_support_locals_st *s = &mode_lib->scratch.dml_core_mode_support_locals; + + CalculateVMRowAndSwath_params->NumberOfActiveSurfaces = mode_lib->ms.num_active_planes; + CalculateVMRowAndSwath_params->myPipe = s->SurfParameters; + CalculateVMRowAndSwath_params->SurfaceSizeInMALL = mode_lib->ms.SurfaceSizeInMALL; + CalculateVMRowAndSwath_params->PTEBufferSizeInRequestsLuma = mode_lib->ms.ip.dpte_buffer_size_in_pte_reqs_luma; + CalculateVMRowAndSwath_params->PTEBufferSizeInRequestsChroma = mode_lib->ms.ip.dpte_buffer_size_in_pte_reqs_chroma; + CalculateVMRowAndSwath_params->DCCMetaBufferSizeBytes = mode_lib->ms.ip.dcc_meta_buffer_size_bytes; + CalculateVMRowAndSwath_params->UseMALLForStaticScreen = mode_lib->ms.cache_display_cfg.plane.UseMALLForStaticScreen; + CalculateVMRowAndSwath_params->UseMALLForPStateChange = mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange; + CalculateVMRowAndSwath_params->MALLAllocatedForDCN = mode_lib->ms.soc.mall_allocated_for_dcn_mbytes; + CalculateVMRowAndSwath_params->SwathWidthY = mode_lib->ms.SwathWidthYThisState; + CalculateVMRowAndSwath_params->SwathWidthC = mode_lib->ms.SwathWidthCThisState; + CalculateVMRowAndSwath_params->GPUVMEnable = mode_lib->ms.cache_display_cfg.plane.GPUVMEnable; + CalculateVMRowAndSwath_params->HostVMEnable = mode_lib->ms.cache_display_cfg.plane.HostVMEnable; + CalculateVMRowAndSwath_params->HostVMMaxNonCachedPageTableLevels = mode_lib->ms.cache_display_cfg.plane.HostVMMaxPageTableLevels; + CalculateVMRowAndSwath_params->GPUVMMaxPageTableLevels = mode_lib->ms.cache_display_cfg.plane.GPUVMMaxPageTableLevels; + CalculateVMRowAndSwath_params->GPUVMMinPageSizeKBytes = mode_lib->ms.cache_display_cfg.plane.GPUVMMinPageSizeKBytes; + CalculateVMRowAndSwath_params->HostVMMinPageSize = mode_lib->ms.soc.hostvm_min_page_size_kbytes * 1024; + CalculateVMRowAndSwath_params->PTEBufferModeOverrideEn = mode_lib->ms.cache_display_cfg.plane.PTEBufferModeOverrideEn; + CalculateVMRowAndSwath_params->PTEBufferModeOverrideVal = mode_lib->ms.cache_display_cfg.plane.PTEBufferMode; + CalculateVMRowAndSwath_params->PTEBufferSizeNotExceeded = mode_lib->ms.PTEBufferSizeNotExceededPerState; + CalculateVMRowAndSwath_params->DCCMetaBufferSizeNotExceeded = mode_lib->ms.DCCMetaBufferSizeNotExceededPerState; + CalculateVMRowAndSwath_params->dpte_row_width_luma_ub = s->dummy_integer_array[0]; + CalculateVMRowAndSwath_params->dpte_row_width_chroma_ub = s->dummy_integer_array[1]; + CalculateVMRowAndSwath_params->dpte_row_height_luma = mode_lib->ms.dpte_row_height; + CalculateVMRowAndSwath_params->dpte_row_height_chroma = mode_lib->ms.dpte_row_height_chroma; + CalculateVMRowAndSwath_params->dpte_row_height_linear_luma = s->dummy_integer_array[2]; // VBA_DELTA + CalculateVMRowAndSwath_params->dpte_row_height_linear_chroma = s->dummy_integer_array[3]; // VBA_DELTA + CalculateVMRowAndSwath_params->meta_req_width = s->dummy_integer_array[4]; + CalculateVMRowAndSwath_params->meta_req_width_chroma = s->dummy_integer_array[5]; + CalculateVMRowAndSwath_params->meta_req_height = s->dummy_integer_array[6]; + CalculateVMRowAndSwath_params->meta_req_height_chroma = s->dummy_integer_array[7]; + CalculateVMRowAndSwath_params->meta_row_width = s->dummy_integer_array[8]; + CalculateVMRowAndSwath_params->meta_row_width_chroma = s->dummy_integer_array[9]; + CalculateVMRowAndSwath_params->meta_row_height = mode_lib->ms.meta_row_height; + CalculateVMRowAndSwath_params->meta_row_height_chroma = mode_lib->ms.meta_row_height_chroma; + CalculateVMRowAndSwath_params->vm_group_bytes = s->dummy_integer_array[10]; + CalculateVMRowAndSwath_params->dpte_group_bytes = mode_lib->ms.dpte_group_bytes; + CalculateVMRowAndSwath_params->PixelPTEReqWidthY = s->dummy_integer_array[11]; + CalculateVMRowAndSwath_params->PixelPTEReqHeightY = s->dummy_integer_array[12]; + CalculateVMRowAndSwath_params->PTERequestSizeY = s->dummy_integer_array[13]; + CalculateVMRowAndSwath_params->PixelPTEReqWidthC = s->dummy_integer_array[14]; + CalculateVMRowAndSwath_params->PixelPTEReqHeightC = s->dummy_integer_array[15]; + CalculateVMRowAndSwath_params->PTERequestSizeC = s->dummy_integer_array[16]; + CalculateVMRowAndSwath_params->dpde0_bytes_per_frame_ub_l = s->dummy_integer_array[17]; + CalculateVMRowAndSwath_params->meta_pte_bytes_per_frame_ub_l = s->dummy_integer_array[18]; + CalculateVMRowAndSwath_params->dpde0_bytes_per_frame_ub_c = s->dummy_integer_array[19]; + CalculateVMRowAndSwath_params->meta_pte_bytes_per_frame_ub_c = s->dummy_integer_array[20]; + CalculateVMRowAndSwath_params->PrefetchSourceLinesY = mode_lib->ms.PrefetchLinesYThisState; + CalculateVMRowAndSwath_params->PrefetchSourceLinesC = mode_lib->ms.PrefetchLinesCThisState; + CalculateVMRowAndSwath_params->VInitPreFillY = mode_lib->ms.PrefillY; + CalculateVMRowAndSwath_params->VInitPreFillC = mode_lib->ms.PrefillC; + CalculateVMRowAndSwath_params->MaxNumSwathY = mode_lib->ms.MaxNumSwY; + CalculateVMRowAndSwath_params->MaxNumSwathC = mode_lib->ms.MaxNumSwC; + CalculateVMRowAndSwath_params->meta_row_bw = mode_lib->ms.meta_row_bandwidth_this_state; + CalculateVMRowAndSwath_params->dpte_row_bw = mode_lib->ms.dpte_row_bandwidth_this_state; + CalculateVMRowAndSwath_params->PixelPTEBytesPerRow = mode_lib->ms.DPTEBytesPerRowThisState; + CalculateVMRowAndSwath_params->PDEAndMetaPTEBytesFrame = mode_lib->ms.PDEAndMetaPTEBytesPerFrameThisState; + CalculateVMRowAndSwath_params->MetaRowByte = mode_lib->ms.MetaRowBytesThisState; + CalculateVMRowAndSwath_params->use_one_row_for_frame = mode_lib->ms.use_one_row_for_frame_this_state; + CalculateVMRowAndSwath_params->use_one_row_for_frame_flip = mode_lib->ms.use_one_row_for_frame_flip_this_state; + CalculateVMRowAndSwath_params->UsesMALLForStaticScreen = s->dummy_boolean_array[0]; + CalculateVMRowAndSwath_params->PTE_BUFFER_MODE = s->dummy_boolean_array[1]; + CalculateVMRowAndSwath_params->BIGK_FRAGMENT_SIZE = s->dummy_integer_array[21]; +} + /// @brief The Mode Support function. dml_bool_t dml_core_mode_support(struct display_mode_lib_st *mode_lib) { @@ -7683,69 +7753,7 @@ dml_bool_t dml_core_mode_support(struct display_mode_lib_st *mode_lib) s->SurfParameters[k].SwathHeightC = mode_lib->ms.SwathHeightCThisState[k]; } - CalculateVMRowAndSwath_params->NumberOfActiveSurfaces = mode_lib->ms.num_active_planes; - CalculateVMRowAndSwath_params->myPipe = s->SurfParameters; - CalculateVMRowAndSwath_params->SurfaceSizeInMALL = mode_lib->ms.SurfaceSizeInMALL; - CalculateVMRowAndSwath_params->PTEBufferSizeInRequestsLuma = mode_lib->ms.ip.dpte_buffer_size_in_pte_reqs_luma; - CalculateVMRowAndSwath_params->PTEBufferSizeInRequestsChroma = mode_lib->ms.ip.dpte_buffer_size_in_pte_reqs_chroma; - CalculateVMRowAndSwath_params->DCCMetaBufferSizeBytes = mode_lib->ms.ip.dcc_meta_buffer_size_bytes; - CalculateVMRowAndSwath_params->UseMALLForStaticScreen = mode_lib->ms.cache_display_cfg.plane.UseMALLForStaticScreen; - CalculateVMRowAndSwath_params->UseMALLForPStateChange = mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange; - CalculateVMRowAndSwath_params->MALLAllocatedForDCN = mode_lib->ms.soc.mall_allocated_for_dcn_mbytes; - CalculateVMRowAndSwath_params->SwathWidthY = mode_lib->ms.SwathWidthYThisState; - CalculateVMRowAndSwath_params->SwathWidthC = mode_lib->ms.SwathWidthCThisState; - CalculateVMRowAndSwath_params->GPUVMEnable = mode_lib->ms.cache_display_cfg.plane.GPUVMEnable; - CalculateVMRowAndSwath_params->HostVMEnable = mode_lib->ms.cache_display_cfg.plane.HostVMEnable; - CalculateVMRowAndSwath_params->HostVMMaxNonCachedPageTableLevels = mode_lib->ms.cache_display_cfg.plane.HostVMMaxPageTableLevels; - CalculateVMRowAndSwath_params->GPUVMMaxPageTableLevels = mode_lib->ms.cache_display_cfg.plane.GPUVMMaxPageTableLevels; - CalculateVMRowAndSwath_params->GPUVMMinPageSizeKBytes = mode_lib->ms.cache_display_cfg.plane.GPUVMMinPageSizeKBytes; - CalculateVMRowAndSwath_params->HostVMMinPageSize = mode_lib->ms.soc.hostvm_min_page_size_kbytes * 1024; - CalculateVMRowAndSwath_params->PTEBufferModeOverrideEn = mode_lib->ms.cache_display_cfg.plane.PTEBufferModeOverrideEn; - CalculateVMRowAndSwath_params->PTEBufferModeOverrideVal = mode_lib->ms.cache_display_cfg.plane.PTEBufferMode; - CalculateVMRowAndSwath_params->PTEBufferSizeNotExceeded = mode_lib->ms.PTEBufferSizeNotExceededPerState; - CalculateVMRowAndSwath_params->DCCMetaBufferSizeNotExceeded = mode_lib->ms.DCCMetaBufferSizeNotExceededPerState; - CalculateVMRowAndSwath_params->dpte_row_width_luma_ub = s->dummy_integer_array[0]; - CalculateVMRowAndSwath_params->dpte_row_width_chroma_ub = s->dummy_integer_array[1]; - CalculateVMRowAndSwath_params->dpte_row_height_luma = mode_lib->ms.dpte_row_height; - CalculateVMRowAndSwath_params->dpte_row_height_chroma = mode_lib->ms.dpte_row_height_chroma; - CalculateVMRowAndSwath_params->dpte_row_height_linear_luma = s->dummy_integer_array[2]; // VBA_DELTA - CalculateVMRowAndSwath_params->dpte_row_height_linear_chroma = s->dummy_integer_array[3]; // VBA_DELTA - CalculateVMRowAndSwath_params->meta_req_width = s->dummy_integer_array[4]; - CalculateVMRowAndSwath_params->meta_req_width_chroma = s->dummy_integer_array[5]; - CalculateVMRowAndSwath_params->meta_req_height = s->dummy_integer_array[6]; - CalculateVMRowAndSwath_params->meta_req_height_chroma = s->dummy_integer_array[7]; - CalculateVMRowAndSwath_params->meta_row_width = s->dummy_integer_array[8]; - CalculateVMRowAndSwath_params->meta_row_width_chroma = s->dummy_integer_array[9]; - CalculateVMRowAndSwath_params->meta_row_height = mode_lib->ms.meta_row_height; - CalculateVMRowAndSwath_params->meta_row_height_chroma = mode_lib->ms.meta_row_height_chroma; - CalculateVMRowAndSwath_params->vm_group_bytes = s->dummy_integer_array[10]; - CalculateVMRowAndSwath_params->dpte_group_bytes = mode_lib->ms.dpte_group_bytes; - CalculateVMRowAndSwath_params->PixelPTEReqWidthY = s->dummy_integer_array[11]; - CalculateVMRowAndSwath_params->PixelPTEReqHeightY = s->dummy_integer_array[12]; - CalculateVMRowAndSwath_params->PTERequestSizeY = s->dummy_integer_array[13]; - CalculateVMRowAndSwath_params->PixelPTEReqWidthC = s->dummy_integer_array[14]; - CalculateVMRowAndSwath_params->PixelPTEReqHeightC = s->dummy_integer_array[15]; - CalculateVMRowAndSwath_params->PTERequestSizeC = s->dummy_integer_array[16]; - CalculateVMRowAndSwath_params->dpde0_bytes_per_frame_ub_l = s->dummy_integer_array[17]; - CalculateVMRowAndSwath_params->meta_pte_bytes_per_frame_ub_l = s->dummy_integer_array[18]; - CalculateVMRowAndSwath_params->dpde0_bytes_per_frame_ub_c = s->dummy_integer_array[19]; - CalculateVMRowAndSwath_params->meta_pte_bytes_per_frame_ub_c = s->dummy_integer_array[20]; - CalculateVMRowAndSwath_params->PrefetchSourceLinesY = mode_lib->ms.PrefetchLinesYThisState; - CalculateVMRowAndSwath_params->PrefetchSourceLinesC = mode_lib->ms.PrefetchLinesCThisState; - CalculateVMRowAndSwath_params->VInitPreFillY = mode_lib->ms.PrefillY; - CalculateVMRowAndSwath_params->VInitPreFillC = mode_lib->ms.PrefillC; - CalculateVMRowAndSwath_params->MaxNumSwathY = mode_lib->ms.MaxNumSwY; - CalculateVMRowAndSwath_params->MaxNumSwathC = mode_lib->ms.MaxNumSwC; - CalculateVMRowAndSwath_params->meta_row_bw = mode_lib->ms.meta_row_bandwidth_this_state; - CalculateVMRowAndSwath_params->dpte_row_bw = mode_lib->ms.dpte_row_bandwidth_this_state; - CalculateVMRowAndSwath_params->PixelPTEBytesPerRow = mode_lib->ms.DPTEBytesPerRowThisState; - CalculateVMRowAndSwath_params->PDEAndMetaPTEBytesFrame = mode_lib->ms.PDEAndMetaPTEBytesPerFrameThisState; - CalculateVMRowAndSwath_params->MetaRowByte = mode_lib->ms.MetaRowBytesThisState; - CalculateVMRowAndSwath_params->use_one_row_for_frame = mode_lib->ms.use_one_row_for_frame_this_state; - CalculateVMRowAndSwath_params->use_one_row_for_frame_flip = mode_lib->ms.use_one_row_for_frame_flip_this_state; - CalculateVMRowAndSwath_params->UsesMALLForStaticScreen = s->dummy_boolean_array[0]; - CalculateVMRowAndSwath_params->PTE_BUFFER_MODE = s->dummy_boolean_array[1]; - CalculateVMRowAndSwath_params->BIGK_FRAGMENT_SIZE = s->dummy_integer_array[21]; + set_vm_row_and_swath_parameters(mode_lib); CalculateVMRowAndSwath(&mode_lib->scratch, CalculateVMRowAndSwath_params); -- cgit From 8fc2796dea6f1210e1a01573961d5836a7ce531e Mon Sep 17 00:00:00 2001 From: Mario Limonciello Date: Fri, 5 Dec 2025 12:41:58 -0600 Subject: drm/amdkfd: Export the cwsr_size and ctl_stack_size to userspace This is important for userspace to avoid hardcoding VGPR size. Reviewed-by: Kent Russell Signed-off-by: Mario Limonciello Signed-off-by: Alex Deucher (cherry picked from commit 71776e0965f9f730af19c5f548827f2a7c91f5a8) Cc: stable@vger.kernel.org --- drivers/gpu/drm/amd/amdkfd/kfd_topology.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_topology.c b/drivers/gpu/drm/amd/amdkfd/kfd_topology.c index 811636af14ea..3eb32d58a120 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_topology.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_topology.c @@ -491,6 +491,10 @@ static ssize_t node_show(struct kobject *kobj, struct attribute *attr, dev->node_props.num_sdma_queues_per_engine); sysfs_show_32bit_prop(buffer, offs, "num_cp_queues", dev->node_props.num_cp_queues); + sysfs_show_32bit_prop(buffer, offs, "cwsr_size", + dev->node_props.cwsr_size); + sysfs_show_32bit_prop(buffer, offs, "ctl_stack_size", + dev->node_props.ctl_stack_size); if (dev->gpu) { log_max_watch_addr = -- cgit From 7329417fc9ac128729c3a092b006c8f1fd0d04a6 Mon Sep 17 00:00:00 2001 From: Rosen Penev Date: Sat, 8 Nov 2025 09:40:47 -0800 Subject: drm/amd/display: shrink struct members On a 32-bit ARM system, the audio_decoder struct ends up being too large for dp_retrain_link_dp_test. link_dp_cts.c:157:1: error: the frame size of 1328 bytes is larger than 1280 bytes [-Werror=frame-larger-than=] This is mitigated by shrinking the members of the struct and avoids having to deal with dynamic allocation. feed_back_divider is assigned but otherwise unused. Remove both. pixel_repetition looks like it should be a bool since it's only ever assigned to 1. But there are checks for 2 and 4. Reduce to uint8_t. Remove ss_percentage_divider. Unused. Shrink refresh_rate as it gets assigned to at most a 3 digit integer value. Signed-off-by: Rosen Penev Reviewed-by: Alex Hung Signed-off-by: Alex Deucher (cherry picked from commit 3849efdc7888d537f09c3dcfaea4b3cd377a102e) --- drivers/gpu/drm/amd/display/dc/hwss/dce110/dce110_hwseq.c | 3 --- drivers/gpu/drm/amd/display/include/audio_types.h | 12 +++++------- 2 files changed, 5 insertions(+), 10 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dce110/dce110_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dce110/dce110_hwseq.c index 8fe399939220..4986f12dc9df 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dce110/dce110_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dce110/dce110_hwseq.c @@ -1484,9 +1484,6 @@ void build_audio_output( state->clk_mgr); } - audio_output->pll_info.feed_back_divider = - pipe_ctx->pll_settings.feedback_divider; - audio_output->pll_info.dto_source = translate_to_dto_source( pipe_ctx->stream_res.tg->inst + 1); diff --git a/drivers/gpu/drm/amd/display/include/audio_types.h b/drivers/gpu/drm/amd/display/include/audio_types.h index e4a26143f14c..6699ad4fa825 100644 --- a/drivers/gpu/drm/amd/display/include/audio_types.h +++ b/drivers/gpu/drm/amd/display/include/audio_types.h @@ -47,15 +47,15 @@ struct audio_crtc_info { uint32_t h_total; uint32_t h_active; uint32_t v_active; - uint32_t pixel_repetition; uint32_t requested_pixel_clock_100Hz; /* in 100Hz */ uint32_t calculated_pixel_clock_100Hz; /* in 100Hz */ - uint32_t refresh_rate; + uint32_t dsc_bits_per_pixel; + uint32_t dsc_num_slices; enum dc_color_depth color_depth; enum dc_pixel_encoding pixel_encoding; + uint16_t refresh_rate; + uint8_t pixel_repetition; bool interlaced; - uint32_t dsc_bits_per_pixel; - uint32_t dsc_num_slices; }; struct azalia_clock_info { uint32_t pixel_clock_in_10khz; @@ -78,11 +78,9 @@ enum audio_dto_source { struct audio_pll_info { uint32_t audio_dto_source_clock_in_khz; - uint32_t feed_back_divider; + uint32_t ss_percentage; enum audio_dto_source dto_source; bool ss_enabled; - uint32_t ss_percentage; - uint32_t ss_percentage_divider; }; struct audio_channel_associate_info { -- cgit From cf326449637a566ba98fb82c47d46cd479608c88 Mon Sep 17 00:00:00 2001 From: Jonathan Kim Date: Fri, 5 Dec 2025 14:41:08 -0500 Subject: drm/amdkfd: bump minimum vgpr size for gfx1151 GFX1151 has 1.5x the number of available physical VGPRs per SIMD. Bump total memory availability for acquire checks on queue creation. Signed-off-by: Jonathan Kim Reviewed-by: Mario Limonciello Signed-off-by: Alex Deucher (cherry picked from commit b42f3bf9536c9b710fd1d4deb7d1b0dc819dc72d) Cc: stable@vger.kernel.org --- drivers/gpu/drm/amd/amdkfd/kfd_queue.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_queue.c b/drivers/gpu/drm/amd/amdkfd/kfd_queue.c index f1e7583650c4..80c4fa2b0975 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_queue.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_queue.c @@ -409,6 +409,7 @@ static u32 kfd_get_vgpr_size_per_cu(u32 gfxv) vgpr_size = 0x80000; else if (gfxv == 110000 || /* GFX_VERSION_PLUM_BONITO */ gfxv == 110001 || /* GFX_VERSION_WHEAT_NAS */ + gfxv == 110501 || /* GFX_VERSION_GFX1151 */ gfxv == 120000 || /* GFX_VERSION_GFX1200 */ gfxv == 120001) /* GFX_VERSION_GFX1201 */ vgpr_size = 0x60000; -- cgit From 2e1da460916626fedbbc8518b9c4e1b064f201ed Mon Sep 17 00:00:00 2001 From: Ivan Lipski Date: Fri, 7 Nov 2025 16:50:37 -0500 Subject: drm/amd/display: Improve HDMI info retrieval [WHY & HOW] Make a dedicated function to read HDMI-related monitor info, including monitor's SCDC support. Fixes: 3471b9a31ce3 ("drm/amd/display: Rework HDMI data channel reads") Suggested-by: Fangzhi Zuo Reviewed-by: Jerry Zuo Signed-off-by: Ivan Lipski Signed-off-by: Alex Hung Signed-off-by: Alex Deucher (cherry picked from commit c78e31bcf586f1c910a2636650840f5ce1cb1c63) --- drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h | 3 +++ drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c | 8 ++++++++ 2 files changed, 11 insertions(+) diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h index ef97cede9926..bd0403005f37 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h @@ -1063,6 +1063,9 @@ int amdgpu_dm_update_plane_color_mgmt(struct dm_crtc_state *crtc, void amdgpu_dm_update_connector_after_detect( struct amdgpu_dm_connector *aconnector); +void populate_hdmi_info_from_connector(struct drm_hdmi_info *info, + struct dc_edid_caps *edid_caps); + extern const struct drm_encoder_helper_funcs amdgpu_dm_encoder_helper_funcs; int amdgpu_dm_process_dmub_aux_transfer_sync(struct dc_context *ctx, unsigned int link_index, diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c index ac98c746c3de..e5e993d3ef74 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c @@ -139,6 +139,9 @@ enum dc_edid_status dm_helpers_parse_edid_caps( edid_caps->edid_hdmi = connector->display_info.is_hdmi; + if (edid_caps->edid_hdmi) + populate_hdmi_info_from_connector(&connector->display_info.hdmi, edid_caps); + apply_edid_quirks(dev, edid_buf, edid_caps); sad_count = drm_edid_to_sad((struct edid *) edid->raw_edid, &sads); @@ -990,6 +993,11 @@ dm_helpers_read_acpi_edid(struct amdgpu_dm_connector *aconnector) return drm_edid_read_custom(connector, dm_helpers_probe_acpi_edid, connector); } +void populate_hdmi_info_from_connector(struct drm_hdmi_info *hdmi, struct dc_edid_caps *edid_caps) +{ + edid_caps->scdc_present = hdmi->scdc.supported; +} + enum dc_edid_status dm_helpers_read_local_edid( struct dc_context *ctx, struct dc_link *link, -- cgit From 460b31720369fc77c23301708641cfa1bf2fcb8f Mon Sep 17 00:00:00 2001 From: Dibin Moolakadan Subrahmanian Date: Tue, 11 Nov 2025 19:24:03 +0530 Subject: drm/i915/fbdev: Hold runtime PM ref during fbdev BO creation MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit During fbdev probe, the xe driver allocates and pins a framebuffer BO (via xe_bo_create_pin_map_novm() → xe_ggtt_insert_bo()). Without a runtime PM reference, xe_pm_runtime_get_noresume() warns about missing outer PM protection as below: xe 0000:03:00.0: [drm] Missing outer runtime PM protection Acquire a runtime PM reference before framebuffer allocation to ensure xe_ggtt_insert_bo() executes under active runtime PM context. Changes in v2: - Update commit message to add Fixes tag (Jani Nikula) Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/6350 Fixes: 44e694958b95 ("drm/xe/display: Implement display support") Cc: Jani Nikula Signed-off-by: Dibin Moolakadan Subrahmanian Reviewed-by: Jouni Högander Signed-off-by: Ankit Nautiyal Link: https://patch.msgid.link/20251111135403.3415947-1-dibin.moolakadan.subrahmanian@intel.com (cherry picked from commit 37fc7b7b3ab0e3bb900657199cd3770a4fda03fb) Signed-off-by: Jani Nikula --- drivers/gpu/drm/i915/display/intel_fbdev.c | 11 +++++++---- 1 file changed, 7 insertions(+), 4 deletions(-) diff --git a/drivers/gpu/drm/i915/display/intel_fbdev.c b/drivers/gpu/drm/i915/display/intel_fbdev.c index 9cd03e2adeb2..44f4fcce526e 100644 --- a/drivers/gpu/drm/i915/display/intel_fbdev.c +++ b/drivers/gpu/drm/i915/display/intel_fbdev.c @@ -288,13 +288,18 @@ int intel_fbdev_driver_fbdev_probe(struct drm_fb_helper *helper, drm_framebuffer_put(&fb->base); fb = NULL; } + + wakeref = intel_display_rpm_get(display); + if (!fb || drm_WARN_ON(display->drm, !intel_fb_bo(&fb->base))) { drm_dbg_kms(display->drm, "no BIOS fb, allocating a new one\n"); fb = __intel_fbdev_fb_alloc(display, sizes); - if (IS_ERR(fb)) - return PTR_ERR(fb); + if (IS_ERR(fb)) { + ret = PTR_ERR(fb); + goto out_unlock; + } } else { drm_dbg_kms(display->drm, "re-using BIOS fb\n"); prealloc = true; @@ -302,8 +307,6 @@ int intel_fbdev_driver_fbdev_probe(struct drm_fb_helper *helper, sizes->fb_height = fb->base.height; } - wakeref = intel_display_rpm_get(display); - /* Pin the GGTT vma for our access via info->screen_base. * This also validates that any existing fb inherited from the * BIOS is suitable for own access. -- cgit From 1c7f9e528f8f488b060b786bfb90b40540854db3 Mon Sep 17 00:00:00 2001 From: Ard Biesheuvel Date: Fri, 5 Dec 2025 12:35:01 +0100 Subject: drm/i915: Fix format string truncation warning MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit GCC notices that the 16-byte uabi_name field could theoretically be too small for the formatted string if the instance number exceeds 100. So grow the field to 20 bytes. drivers/gpu/drm/i915/intel_memory_region.c: In function ‘intel_memory_region_create’: drivers/gpu/drm/i915/intel_memory_region.c:273:61: error: ‘%u’ directive output may be truncated writing between 1 and 5 bytes into a region of size between 3 and 11 [-Werror=format-truncation=] 273 | snprintf(mem->uabi_name, sizeof(mem->uabi_name), "%s%u", | ^~ drivers/gpu/drm/i915/intel_memory_region.c:273:58: note: directive argument in the range [0, 65535] 273 | snprintf(mem->uabi_name, sizeof(mem->uabi_name), "%s%u", | ^~~~~~ drivers/gpu/drm/i915/intel_memory_region.c:273:9: note: ‘snprintf’ output between 7 and 19 bytes into a destination of size 16 273 | snprintf(mem->uabi_name, sizeof(mem->uabi_name), "%s%u", | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 274 | intel_memory_type_str(type), instance); | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Fixes: 3b38d3515753 ("drm/i915: Add stable memory region names") Cc: # v6.8+ Signed-off-by: Ard Biesheuvel Signed-off-by: Tvrtko Ursulin Link: https://lore.kernel.org/r/20251205113500.684286-2-ardb@kernel.org (cherry picked from commit 18476087f1a18dc279d200d934ad94fba1fb51d5) Signed-off-by: Jani Nikula --- drivers/gpu/drm/i915/intel_memory_region.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/i915/intel_memory_region.h b/drivers/gpu/drm/i915/intel_memory_region.h index b3b75be9ced5..e9a4e6090fe0 100644 --- a/drivers/gpu/drm/i915/intel_memory_region.h +++ b/drivers/gpu/drm/i915/intel_memory_region.h @@ -72,7 +72,7 @@ struct intel_memory_region { u16 instance; enum intel_region_id id; char name[16]; - char uabi_name[16]; + char uabi_name[20]; bool private; /* not for userspace */ struct { -- cgit From 93a01629c8bfd30906c76921ec986802d76920c6 Mon Sep 17 00:00:00 2001 From: "Mario Limonciello (AMD)" Date: Mon, 8 Dec 2025 22:46:46 -0600 Subject: drm/amd: Fix unbind/rebind for VCN 4.0.5 Unbinding amdgpu has no problems, but binding it again leads to an error of sysfs file already existing. This is because it wasn't actually cleaned up on unbind. Add the missing cleanup step. Fixes: 547aad32edac ("drm/amdgpu: add VCN4 ip block support") Signed-off-by: Mario Limonciello (AMD) Reviewed-by: Alex Deucher Signed-off-by: Alex Deucher (cherry picked from commit d717e62e9b6ccff0e3cec78a58dfbd00858448b3) Cc: stable@vger.kernel.org --- drivers/gpu/drm/amd/amdgpu/vcn_v4_0_5.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_5.c b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_5.c index b107ee80e472..1f6a22983c0d 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_5.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_5.c @@ -265,6 +265,8 @@ static int vcn_v4_0_5_sw_fini(struct amdgpu_ip_block *ip_block) if (amdgpu_sriov_vf(adev)) amdgpu_virt_free_mm_table(adev); + amdgpu_vcn_sysfs_reset_mask_fini(adev); + for (i = 0; i < adev->vcn.num_vcn_inst; i++) { r = amdgpu_vcn_suspend(adev, i); if (r) -- cgit From 72e24456a54fe04710d89626cc5a88703e2f6202 Mon Sep 17 00:00:00 2001 From: Mario Limonciello Date: Tue, 9 Dec 2025 11:14:47 -0600 Subject: Revert "drm/amd/display: Fix pbn to kbps Conversion" Deeply daisy chained DP/MST displays are no longer able to light up. This reverts commit e0dec00f3d05 ("drm/amd/display: Fix pbn to kbps Conversion") Cc: Jerry Zuo Reported-by: nat@nullable.se Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/4756 Signed-off-by: Mario Limonciello Acked-by: Alex Deucher Signed-off-by: Alex Deucher (cherry picked from commit e1c94109c76e8a77a21531bd53f6c63356c81158) Cc: stable@vger.kernel.org # 6.17+ --- .../amd/display/amdgpu_dm/amdgpu_dm_mst_types.c | 59 +++++++++++++--------- 1 file changed, 36 insertions(+), 23 deletions(-) diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c index dbd1da4d85d3..5e92eaa67aa3 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c @@ -884,28 +884,26 @@ struct dsc_mst_fairness_params { }; #if defined(CONFIG_DRM_AMD_DC_FP) -static uint64_t kbps_to_pbn(int kbps, bool is_peak_pbn) +static uint16_t get_fec_overhead_multiplier(struct dc_link *dc_link) { - uint64_t effective_kbps = (uint64_t)kbps; + u8 link_coding_cap; + uint16_t fec_overhead_multiplier_x1000 = PBN_FEC_OVERHEAD_MULTIPLIER_8B_10B; - if (is_peak_pbn) { // add 0.6% (1006/1000) overhead into effective kbps - effective_kbps *= 1006; - effective_kbps = div_u64(effective_kbps, 1000); - } + link_coding_cap = dc_link_dp_mst_decide_link_encoding_format(dc_link); + if (link_coding_cap == DP_128b_132b_ENCODING) + fec_overhead_multiplier_x1000 = PBN_FEC_OVERHEAD_MULTIPLIER_128B_132B; - return (uint64_t) DIV64_U64_ROUND_UP(effective_kbps * 64, (54 * 8 * 1000)); + return fec_overhead_multiplier_x1000; } -static uint32_t pbn_to_kbps(unsigned int pbn, bool with_margin) +static int kbps_to_peak_pbn(int kbps, uint16_t fec_overhead_multiplier_x1000) { - uint64_t pbn_effective = (uint64_t)pbn; - - if (with_margin) // deduct 0.6% (994/1000) overhead from effective pbn - pbn_effective *= (1000000 / PEAK_FACTOR_X1000); - else - pbn_effective *= 1000; + u64 peak_kbps = kbps; - return DIV_U64_ROUND_UP(pbn_effective * 8 * 54, 64); + peak_kbps *= 1006; + peak_kbps *= fec_overhead_multiplier_x1000; + peak_kbps = div_u64(peak_kbps, 1000 * 1000); + return (int) DIV64_U64_ROUND_UP(peak_kbps * 64, (54 * 8 * 1000)); } static void set_dsc_configs_from_fairness_vars(struct dsc_mst_fairness_params *params, @@ -976,7 +974,7 @@ static int bpp_x16_from_pbn(struct dsc_mst_fairness_params param, int pbn) dc_dsc_get_default_config_option(param.sink->ctx->dc, &dsc_options); dsc_options.max_target_bpp_limit_override_x16 = drm_connector->display_info.max_dsc_bpp * 16; - kbps = pbn_to_kbps(pbn, false); + kbps = div_u64((u64)pbn * 994 * 8 * 54, 64); dc_dsc_compute_config( param.sink->ctx->dc->res_pool->dscs[0], ¶m.sink->dsc_caps.dsc_dec_caps, @@ -1005,11 +1003,12 @@ static int increase_dsc_bpp(struct drm_atomic_state *state, int link_timeslots_used; int fair_pbn_alloc; int ret = 0; + uint16_t fec_overhead_multiplier_x1000 = get_fec_overhead_multiplier(dc_link); for (i = 0; i < count; i++) { if (vars[i + k].dsc_enabled) { initial_slack[i] = - kbps_to_pbn(params[i].bw_range.max_kbps, false) - vars[i + k].pbn; + kbps_to_peak_pbn(params[i].bw_range.max_kbps, fec_overhead_multiplier_x1000) - vars[i + k].pbn; bpp_increased[i] = false; remaining_to_increase += 1; } else { @@ -1105,6 +1104,7 @@ static int try_disable_dsc(struct drm_atomic_state *state, int next_index; int remaining_to_try = 0; int ret; + uint16_t fec_overhead_multiplier_x1000 = get_fec_overhead_multiplier(dc_link); int var_pbn; for (i = 0; i < count; i++) { @@ -1137,7 +1137,7 @@ static int try_disable_dsc(struct drm_atomic_state *state, DRM_DEBUG_DRIVER("MST_DSC index #%d, try no compression\n", next_index); var_pbn = vars[next_index].pbn; - vars[next_index].pbn = kbps_to_pbn(params[next_index].bw_range.stream_kbps, true); + vars[next_index].pbn = kbps_to_peak_pbn(params[next_index].bw_range.stream_kbps, fec_overhead_multiplier_x1000); ret = drm_dp_atomic_find_time_slots(state, params[next_index].port->mgr, params[next_index].port, @@ -1197,6 +1197,7 @@ static int compute_mst_dsc_configs_for_link(struct drm_atomic_state *state, int count = 0; int i, k, ret; bool debugfs_overwrite = false; + uint16_t fec_overhead_multiplier_x1000 = get_fec_overhead_multiplier(dc_link); struct drm_connector_state *new_conn_state; memset(params, 0, sizeof(params)); @@ -1277,7 +1278,7 @@ static int compute_mst_dsc_configs_for_link(struct drm_atomic_state *state, DRM_DEBUG_DRIVER("MST_DSC Try no compression\n"); for (i = 0; i < count; i++) { vars[i + k].aconnector = params[i].aconnector; - vars[i + k].pbn = kbps_to_pbn(params[i].bw_range.stream_kbps, false); + vars[i + k].pbn = kbps_to_peak_pbn(params[i].bw_range.stream_kbps, fec_overhead_multiplier_x1000); vars[i + k].dsc_enabled = false; vars[i + k].bpp_x16 = 0; ret = drm_dp_atomic_find_time_slots(state, params[i].port->mgr, params[i].port, @@ -1299,7 +1300,7 @@ static int compute_mst_dsc_configs_for_link(struct drm_atomic_state *state, DRM_DEBUG_DRIVER("MST_DSC Try max compression\n"); for (i = 0; i < count; i++) { if (params[i].compression_possible && params[i].clock_force_enable != DSC_CLK_FORCE_DISABLE) { - vars[i + k].pbn = kbps_to_pbn(params[i].bw_range.min_kbps, false); + vars[i + k].pbn = kbps_to_peak_pbn(params[i].bw_range.min_kbps, fec_overhead_multiplier_x1000); vars[i + k].dsc_enabled = true; vars[i + k].bpp_x16 = params[i].bw_range.min_target_bpp_x16; ret = drm_dp_atomic_find_time_slots(state, params[i].port->mgr, @@ -1307,7 +1308,7 @@ static int compute_mst_dsc_configs_for_link(struct drm_atomic_state *state, if (ret < 0) return ret; } else { - vars[i + k].pbn = kbps_to_pbn(params[i].bw_range.stream_kbps, false); + vars[i + k].pbn = kbps_to_peak_pbn(params[i].bw_range.stream_kbps, fec_overhead_multiplier_x1000); vars[i + k].dsc_enabled = false; vars[i + k].bpp_x16 = 0; ret = drm_dp_atomic_find_time_slots(state, params[i].port->mgr, @@ -1762,6 +1763,18 @@ clean_exit: return ret; } +static uint32_t kbps_from_pbn(unsigned int pbn) +{ + uint64_t kbps = (uint64_t)pbn; + + kbps *= (1000000 / PEAK_FACTOR_X1000); + kbps *= 8; + kbps *= 54; + kbps /= 64; + + return (uint32_t)kbps; +} + static bool is_dsc_common_config_possible(struct dc_stream_state *stream, struct dc_dsc_bw_range *bw_range) { @@ -1860,7 +1873,7 @@ enum dc_status dm_dp_mst_is_port_support_mode( dc_link_get_highest_encoding_format(stream->link)); cur_link_settings = stream->link->verified_link_cap; root_link_bw_in_kbps = dc_link_bandwidth_kbps(aconnector->dc_link, &cur_link_settings); - virtual_channel_bw_in_kbps = pbn_to_kbps(aconnector->mst_output_port->full_pbn, true); + virtual_channel_bw_in_kbps = kbps_from_pbn(aconnector->mst_output_port->full_pbn); /* pick the end to end bw bottleneck */ end_to_end_bw_in_kbps = min(root_link_bw_in_kbps, virtual_channel_bw_in_kbps); @@ -1913,7 +1926,7 @@ enum dc_status dm_dp_mst_is_port_support_mode( immediate_upstream_port = aconnector->mst_output_port->parent->port_parent; if (immediate_upstream_port) { - virtual_channel_bw_in_kbps = pbn_to_kbps(immediate_upstream_port->full_pbn, true); + virtual_channel_bw_in_kbps = kbps_from_pbn(immediate_upstream_port->full_pbn); virtual_channel_bw_in_kbps = min(root_link_bw_in_kbps, virtual_channel_bw_in_kbps); } else { /* For topology LCT 1 case - only one mstb*/ -- cgit