From: Huacai Chen <chenhc(a)lemote.com>
commit f4d3da72a76a9ce5f57bba64788931686a9dc333 upstream.
In Mesa, dev_info.gart_page_size is used for alignment and it was
set to AMDGPU_GPU_PAGE_SIZE(4KB). However, the page table of AMDGPU
driver requires an alignment on CPU pages. So, for non-4KB page system,
gart_page_size should be max_t(u32, PAGE_SIZE, AMDGPU_GPU_PAGE_SIZE).
Signed-off-by: Rui Wang <wangr(a)lemote.com>
Signed-off-by: Huacai Chen <chenhc(a)lemote.com>
Link: https://github.com/loongson-community/linux-stable/commit/caa9c0a1
[Xi: rebased for drm-next, use max_t for checkpatch,
and reworded commit message.]
Signed-off-by: Xi Ruoyao <xry111(a)mengyan1223.wang>
BugLink: https://gitlab.freedesktop.org/drm/amd/-/issues/1549
Tested-by: Dan Horák <dan(a)danny.cz>
Reviewed-by: Christian König <christian.koenig(a)amd.com>
Signed-off-by: Alex Deucher <alexander.deucher(a)amd.com>
[Salvatore Bonaccorso: Backport to 5.10.y which does not contain
a5a52a43eac0 ("drm/amd/amdgpu/amdgpu_kms: Remove 'struct
drm_amdgpu_info_device dev_info' from the stack") which removes dev_info
from the stack and places it on the heap.]
Tested-by: Timothy Pearson <tpearson(a)raptorengineering.com>
Signed-off-by: Salvatore Bonaccorso <carnil(a)debian.org>
---
drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
index efda38349a03..917b94002f4b 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
@@ -766,9 +766,9 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file
dev_info.high_va_offset = AMDGPU_GMC_HOLE_END;
dev_info.high_va_max = AMDGPU_GMC_HOLE_END | vm_size;
}
- dev_info.virtual_address_alignment = max((int)PAGE_SIZE, AMDGPU_GPU_PAGE_SIZE);
+ dev_info.virtual_address_alignment = max_t(u32, PAGE_SIZE, AMDGPU_GPU_PAGE_SIZE);
dev_info.pte_fragment_size = (1 << adev->vm_manager.fragment_size) * AMDGPU_GPU_PAGE_SIZE;
- dev_info.gart_page_size = AMDGPU_GPU_PAGE_SIZE;
+ dev_info.gart_page_size = max_t(u32, PAGE_SIZE, AMDGPU_GPU_PAGE_SIZE);
dev_info.cu_active_number = adev->gfx.cu_info.number;
dev_info.cu_ao_mask = adev->gfx.cu_info.ao_cu_mask;
dev_info.ce_ram_size = adev->gfx.ce_ram_size;
--
2.34.1
There is a hung_task issue with running generic/068 on an SMR
device. The hang occurs while a process is trying to thaw the
filesystem. The process is trying to take sb->s_umount to thaw the
FS. The lock is held by fsstress, which calls btrfs_sync_fs() and is
waiting for an ordered extent to finish. However, as the FS is frozen,
the ordered extent never finish.
Having an ordered extent while the FS is frozen is the root cause of
the hang. The ordered extent is initiated from btrfs_relocate_chunk()
which is called from btrfs_reclaim_bgs_work().
This commit add sb_*_write() around btrfs_relocate_chunk() call
site. For the usual "btrfs balance" command, we already call it with
mnt_want_file() in btrfs_ioctl_balance().
Additionally, add an ASSERT in btrfs_relocate_chunk() to check it is
properly called.
Fixes: 18bb8bbf13c1 ("btrfs: zoned: automatically reclaim zones")
Cc: stable(a)vger.kernel.org # 5.13+
Link: https://github.com/naota/linux/issues/56
Signed-off-by: Naohiro Aota <naohiro.aota(a)wdc.com>
---
fs/btrfs/block-group.c | 8 +++++++-
fs/btrfs/volumes.c | 6 ++++++
2 files changed, 13 insertions(+), 1 deletion(-)
diff --git a/fs/btrfs/block-group.c b/fs/btrfs/block-group.c
index 68feabc83a27..f9ca58a90630 100644
--- a/fs/btrfs/block-group.c
+++ b/fs/btrfs/block-group.c
@@ -1513,8 +1513,12 @@ void btrfs_reclaim_bgs_work(struct work_struct *work)
if (!test_bit(BTRFS_FS_OPEN, &fs_info->flags))
return;
- if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_BALANCE))
+ sb_start_write(fs_info->sb);
+
+ if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_BALANCE)) {
+ sb_end_write(fs_info->sb);
return;
+ }
/*
* Long running balances can keep us blocked here for eternity, so
@@ -1522,6 +1526,7 @@ void btrfs_reclaim_bgs_work(struct work_struct *work)
*/
if (!mutex_trylock(&fs_info->reclaim_bgs_lock)) {
btrfs_exclop_finish(fs_info);
+ sb_end_write(fs_info->sb);
return;
}
@@ -1596,6 +1601,7 @@ void btrfs_reclaim_bgs_work(struct work_struct *work)
spin_unlock(&fs_info->unused_bgs_lock);
mutex_unlock(&fs_info->reclaim_bgs_lock);
btrfs_exclop_finish(fs_info);
+ sb_end_write(fs_info->sb);
}
void btrfs_reclaim_bgs(struct btrfs_fs_info *fs_info)
diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index b07d382d53a8..e415fb4a698e 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -3251,6 +3251,9 @@ int btrfs_relocate_chunk(struct btrfs_fs_info *fs_info, u64 chunk_offset)
u64 length;
int ret;
+ /* Assert we called sb_start_write(), not to race with FS freezing */
+ sb_assert_write_started(fs_info->sb);
+
/*
* Prevent races with automatic removal of unused block groups.
* After we relocate and before we remove the chunk with offset
@@ -8299,10 +8302,12 @@ static int relocating_repair_kthread(void *data)
target = cache->start;
btrfs_put_block_group(cache);
+ sb_start_write(fs_info->sb);
if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_BALANCE)) {
btrfs_info(fs_info,
"zoned: skip relocating block group %llu to repair: EBUSY",
target);
+ sb_end_write(fs_info->sb);
return -EBUSY;
}
@@ -8330,6 +8335,7 @@ static int relocating_repair_kthread(void *data)
btrfs_put_block_group(cache);
mutex_unlock(&fs_info->reclaim_bgs_lock);
btrfs_exclop_finish(fs_info);
+ sb_end_write(fs_info->sb);
return ret;
}
--
2.35.1