From 5cd5a49c7213315d83bca54a2eee97ba57d256cc Mon Sep 17 00:00:00 2001 From: Shaun Tancheff Date: Fri, 2 Dec 2022 04:19:59 -0600 Subject: [PATCH 1/1] LU-16321 osd: Allow fiemap on kernel buffers Linux commit v5.17-rc3-19-g967747bbc084 uaccess: remove CONFIG_SET_FS When KERNEL_DS gone lustre needs an alternative for fiemap to copy extents to kernel space memory. Direct in-kernel calls to inode->f_ops->fiemap() can utilize an otherwise unused flag on fiemap_extent_info fi_flags to indicate the fiemap extent buffer is allocated in kernel space. Include ldiskfs patches for ldiskfs_fiemap() to define EXT4_FIEMAP_FLAG_MEMCPY and utilize it. HPE-bug-id: LUS-11337 Fixes: d0337cab8e ("LU-14195 osd: don't use set_fs() for ->fiemap() calls.") Signed-off-by: Shaun Tancheff Change-Id: I7a8edb481833fd1bdcf7b6cd6e08397c1754baee Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/49190 Tested-by: jenkins Tested-by: Maloo Reviewed-by: Neil Brown Reviewed-by: Alexander Boyko Reviewed-by: Petros Koutoupis Reviewed-by: Oleg Drokin --- .../linux-5.10/ext4-fiemap-kernel-data.patch | 312 +++++++++++++++++++++ .../kernel_patches/series/ldiskfs-5.10.0-ml.series | 1 + .../series/ldiskfs-5.11.0-40-ubuntu20.series | 1 + .../series/ldiskfs-5.14-rhel9.series | 1 + .../series/ldiskfs-5.14.21-sles15sp4.series | 1 + lustre/osd-ldiskfs/osd_io.c | 23 +- 6 files changed, 338 insertions(+), 1 deletion(-) create mode 100644 ldiskfs/kernel_patches/patches/linux-5.10/ext4-fiemap-kernel-data.patch diff --git a/ldiskfs/kernel_patches/patches/linux-5.10/ext4-fiemap-kernel-data.patch b/ldiskfs/kernel_patches/patches/linux-5.10/ext4-fiemap-kernel-data.patch new file mode 100644 index 0000000..4f7cbc9 --- /dev/null +++ b/ldiskfs/kernel_patches/patches/linux-5.10/ext4-fiemap-kernel-data.patch @@ -0,0 +1,312 @@ +Subject: [PATCH] ext4-fiemap-kernel-data + +Pull in enough upstream fiemap handling to conditionally use +memcpy instead of copy_to_user in fiemap_fill_next_extent. +Common kernel functions prefixed with ext4_ or _ext4_ + +--- + fs/ext4/ext4.h | 3 + + fs/ext4/extents.c | 265 +++++++++++++++++++++++++++++++++++++++++++++- + 2 files changed, 264 insertions(+), 4 deletions(-) + +diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h +index c931e3a..0bb54fe 100644 +--- a/fs/ext4/ext4.h ++++ b/fs/ext4/ext4.h +@@ -766,6 +766,9 @@ enum { + */ + #define EXT4_FIEMAP_EXTENT_HOLE 0x08000000 + ++/* Otherwise unused fi_flags ext4 use memcpy instead of copy_[to|from]_uiser */ ++#define EXT4_FIEMAP_FLAG_MEMCPY 0x80000000 ++ + /* Max physical block we can address w/o extents */ + #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF + +diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c +index 2e62f83..176d2b8 100644 +--- a/fs/ext4/extents.c ++++ b/fs/ext4/extents.c +@@ -2138,6 +2138,264 @@ cleanup: + return err; + } + ++#ifdef KERNEL_DS ++#define ext4_iomap_fiemap(i, f, s, l, ops) \ ++ iomap_fiemap((i), (f), (s), (l), (ops)) ++#else ++/* ++ * linux: ++ * ext4_fiemap_fill_next_extent <--- fiemap_fill_next_extent ++ * ext4_iomap_to_fiemap <----------- iomap_to_fiemap ++ * ext4_iomap_fiemap_actor <-------- iomap_fiemap_actor ++ * ext4_iomap_apply <----------------iomap_apply ++ * _ext4_iomap_fiemap <------------- iomap_fiemap ++ */ ++/** ++ * ext4_fiemap_fill_next_extent - Fiemap helper function ++ * @fieinfo: Fiemap context passed into ->fiemap ++ * @logical: Extent logical start offset, in bytes ++ * @phys: Extent physical start offset, in bytes ++ * @len: Extent length, in bytes ++ * @flags: FIEMAP_EXTENT flags that describe this extent ++ * ++ * Called from file system ->fiemap callback. Will populate extent ++ * info as passed in via arguments and copy to user memory. On ++ * success, extent count on fieinfo is incremented. ++ * ++ * Returns 0 on success, -errno on error, 1 if this was the last ++ * extent that will fit in user array. ++ */ ++#define SET_UNKNOWN_FLAGS (FIEMAP_EXTENT_DELALLOC) ++#define SET_NO_UNMOUNTED_IO_FLAGS (FIEMAP_EXTENT_DATA_ENCRYPTED) ++#define SET_NOT_ALIGNED_FLAGS (FIEMAP_EXTENT_DATA_TAIL|FIEMAP_EXTENT_DATA_INLINE) ++static int ext4_fiemap_fill_next_extent(struct fiemap_extent_info *fieinfo, ++ u64 logical, u64 phys, u64 len, ++ u32 flags) ++{ ++ struct fiemap_extent extent; ++ struct fiemap_extent __user *dest = fieinfo->fi_extents_start; ++ ++ /* only count the extents */ ++ if (fieinfo->fi_extents_max == 0) { ++ fieinfo->fi_extents_mapped++; ++ return (flags & FIEMAP_EXTENT_LAST) ? 1 : 0; ++ } ++ ++ if (fieinfo->fi_extents_mapped >= fieinfo->fi_extents_max) ++ return 1; ++ ++ if (flags & SET_UNKNOWN_FLAGS) ++ flags |= FIEMAP_EXTENT_UNKNOWN; ++ if (flags & SET_NO_UNMOUNTED_IO_FLAGS) ++ flags |= FIEMAP_EXTENT_ENCODED; ++ if (flags & SET_NOT_ALIGNED_FLAGS) ++ flags |= FIEMAP_EXTENT_NOT_ALIGNED; ++ ++ memset(&extent, 0, sizeof(extent)); ++ extent.fe_logical = logical; ++ extent.fe_physical = phys; ++ extent.fe_length = len; ++ extent.fe_flags = flags; ++ ++ dest += fieinfo->fi_extents_mapped; ++ if (fieinfo->fi_flags & EXT4_FIEMAP_FLAG_MEMCPY) ++ memcpy((__force void *)dest, &extent, sizeof(extent)); ++ else if (copy_to_user(dest, &extent, sizeof(extent))) ++ return -EFAULT; ++ ++ fieinfo->fi_extents_mapped++; ++ if (fieinfo->fi_extents_mapped == fieinfo->fi_extents_max) ++ return 1; ++ return (flags & FIEMAP_EXTENT_LAST) ? 1 : 0; ++} ++ ++static int ext4_iomap_to_fiemap(struct fiemap_extent_info *fi, ++ struct iomap *iomap, u32 flags) ++{ ++ switch (iomap->type) { ++ case IOMAP_HOLE: ++ /* skip holes */ ++ return 0; ++ case IOMAP_DELALLOC: ++ flags |= FIEMAP_EXTENT_DELALLOC | FIEMAP_EXTENT_UNKNOWN; ++ break; ++ case IOMAP_MAPPED: ++ break; ++ case IOMAP_UNWRITTEN: ++ flags |= FIEMAP_EXTENT_UNWRITTEN; ++ break; ++ case IOMAP_INLINE: ++ flags |= FIEMAP_EXTENT_DATA_INLINE; ++ break; ++ } ++ ++ if (iomap->flags & IOMAP_F_MERGED) ++ flags |= FIEMAP_EXTENT_MERGED; ++ if (iomap->flags & IOMAP_F_SHARED) ++ flags |= FIEMAP_EXTENT_SHARED; ++ ++ return ext4_fiemap_fill_next_extent(fi, iomap->offset, ++ iomap->addr != IOMAP_NULL_ADDR ? iomap->addr : 0, ++ iomap->length, flags); ++} ++ ++struct fiemap_ctx { ++ struct fiemap_extent_info *fi; ++ struct iomap prev; ++}; ++ ++static loff_t ++ext4_iomap_fiemap_actor(struct inode *inode, loff_t pos, loff_t length, ++ void *data, struct iomap *iomap, struct iomap *srcmap) ++{ ++ struct fiemap_ctx *ctx = data; ++ loff_t ret = length; ++ ++ if (iomap->type == IOMAP_HOLE) ++ return length; ++ ++ ret = ext4_iomap_to_fiemap(ctx->fi, &ctx->prev, 0); ++ ctx->prev = *iomap; ++ switch (ret) { ++ case 0: /* success */ ++ return length; ++ case 1: /* extent array full */ ++ return 0; ++ default: ++ return ret; ++ } ++} ++ ++/* ++ * Execute a iomap write on a segment of the mapping that spans a ++ * contiguous range of pages that have identical block mapping state. ++ * ++ * This avoids the need to map pages individually, do individual allocations ++ * for each page and most importantly avoid the need for filesystem specific ++ * locking per page. Instead, all the operations are amortised over the entire ++ * range of pages. It is assumed that the filesystems will lock whatever ++ * resources they require in the iomap_begin call, and release them in the ++ * iomap_end call. ++ */ ++static loff_t ++ext4_iomap_apply(struct inode *inode, loff_t pos, loff_t length, unsigned flags, ++ const struct iomap_ops *ops, void *data, iomap_actor_t actor) ++{ ++ struct iomap iomap = { .type = IOMAP_HOLE }; ++ struct iomap srcmap = { .type = IOMAP_HOLE }; ++ loff_t written = 0, ret; ++ u64 end; ++ ++ /* ++ * Need to map a range from start position for length bytes. This can ++ * span multiple pages - it is only guaranteed to return a range of a ++ * single type of pages (e.g. all into a hole, all mapped or all ++ * unwritten). Failure at this point has nothing to undo. ++ * ++ * If allocation is required for this range, reserve the space now so ++ * that the allocation is guaranteed to succeed later on. Once we copy ++ * the data into the page cache pages, then we cannot fail otherwise we ++ * expose transient stale data. If the reserve fails, we can safely ++ * back out at this point as there is nothing to undo. ++ */ ++ ret = ops->iomap_begin(inode, pos, length, flags, &iomap, &srcmap); ++ if (ret) ++ return ret; ++ if (WARN_ON(iomap.offset > pos)) { ++ written = -EIO; ++ goto out; ++ } ++ if (WARN_ON(iomap.length == 0)) { ++ written = -EIO; ++ goto out; ++ } ++ ++ /* ++ * Cut down the length to the one actually provided by the filesystem, ++ * as it might not be able to give us the whole size that we requested. ++ */ ++ end = iomap.offset + iomap.length; ++ if (srcmap.type != IOMAP_HOLE) ++ end = min(end, srcmap.offset + srcmap.length); ++ if (pos + length > end) ++ length = end - pos; ++ ++ /* ++ * Now that we have guaranteed that the space allocation will succeed, ++ * we can do the copy-in page by page without having to worry about ++ * failures exposing transient data. ++ * ++ * To support COW operations, we read in data for partially blocks from ++ * the srcmap if the file system filled it in. In that case we the ++ * length needs to be limited to the earlier of the ends of the iomaps. ++ * If the file system did not provide a srcmap we pass in the normal ++ * iomap into the actors so that they don't need to have special ++ * handling for the two cases. ++ */ ++ written = actor(inode, pos, length, data, &iomap, ++ srcmap.type != IOMAP_HOLE ? &srcmap : &iomap); ++ ++out: ++ /* ++ * Now the data has been copied, commit the range we've copied. This ++ * should not fail unless the filesystem has had a fatal error. ++ */ ++ if (ops->iomap_end) { ++ ret = ops->iomap_end(inode, pos, length, ++ written > 0 ? written : 0, ++ flags, &iomap); ++ } ++ ++ return written ? written : ret; ++} ++ ++static ++int _ext4_iomap_fiemap(struct inode *inode, struct fiemap_extent_info *fi, ++ u64 start, u64 len, const struct iomap_ops *ops) ++{ ++ struct fiemap_ctx ctx; ++ loff_t ret; ++ bool in_kernel = fi->fi_flags & EXT4_FIEMAP_FLAG_MEMCPY; ++ ++ memset(&ctx, 0, sizeof(ctx)); ++ ctx.fi = fi; ++ ctx.prev.type = IOMAP_HOLE; ++ ++ fi->fi_flags &= ~EXT4_FIEMAP_FLAG_MEMCPY; ++ ret = fiemap_prep(inode, fi, start, &len, 0); ++ if (in_kernel) ++ fi->fi_flags |= EXT4_FIEMAP_FLAG_MEMCPY; ++ if (ret) ++ return ret; ++ ++ while (len > 0) { ++ ret = ext4_iomap_apply(inode, start, len, IOMAP_REPORT, ops, ++ &ctx, ext4_iomap_fiemap_actor); ++ /* inode with no (attribute) mapping will give ENOENT */ ++ if (ret == -ENOENT) ++ break; ++ if (ret < 0) ++ return ret; ++ if (ret == 0) ++ break; ++ ++ start += ret; ++ len -= ret; ++ } ++ ++ if (ctx.prev.type != IOMAP_HOLE) { ++ ret = ext4_iomap_to_fiemap(fi, &ctx.prev, FIEMAP_EXTENT_LAST); ++ if (ret < 0) ++ return ret; ++ } ++ ++ return 0; ++} ++ ++#define ext4_iomap_fiemap(i, f, s, l, ops) \ ++ _ext4_iomap_fiemap((i), (f), (s), (l), (ops)) ++#endif /* KERNEL_DS */ ++ + static int ext4_fill_es_cache_info(struct inode *inode, + ext4_lblk_t block, ext4_lblk_t num, + struct fiemap_extent_info *fieinfo) +@@ -4918,11 +5176,10 @@ int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, + + if (fieinfo->fi_flags & FIEMAP_FLAG_XATTR) { + fieinfo->fi_flags &= ~FIEMAP_FLAG_XATTR; +- return iomap_fiemap(inode, fieinfo, start, len, +- &ext4_iomap_xattr_ops); ++ return ext4_iomap_fiemap(inode, fieinfo, start, len, ++ &ext4_iomap_xattr_ops); + } +- +- return iomap_fiemap(inode, fieinfo, start, len, &ext4_iomap_report_ops); ++ return ext4_iomap_fiemap(inode, fieinfo, start, len, &ext4_iomap_report_ops); + } + + int ext4_get_es_cache(struct inode *inode, struct fiemap_extent_info *fieinfo, +-- +2.34.1 + diff --git a/ldiskfs/kernel_patches/series/ldiskfs-5.10.0-ml.series b/ldiskfs/kernel_patches/series/ldiskfs-5.10.0-ml.series index 6e2f0dd..fba982a 100644 --- a/ldiskfs/kernel_patches/series/ldiskfs-5.10.0-ml.series +++ b/ldiskfs/kernel_patches/series/ldiskfs-5.10.0-ml.series @@ -30,3 +30,4 @@ rhel8/ext4-ialloc-uid-gid-and-pass-owner-down.patch base/ext4-projid-xattrs.patch linux-5.8/ext4-enc-flag.patch base/ext4-delayed-iput.patch +linux-5.10/ext4-fiemap-kernel-data.patch diff --git a/ldiskfs/kernel_patches/series/ldiskfs-5.11.0-40-ubuntu20.series b/ldiskfs/kernel_patches/series/ldiskfs-5.11.0-40-ubuntu20.series index 4790c81..36705f7 100644 --- a/ldiskfs/kernel_patches/series/ldiskfs-5.11.0-40-ubuntu20.series +++ b/ldiskfs/kernel_patches/series/ldiskfs-5.11.0-40-ubuntu20.series @@ -30,3 +30,4 @@ rhel8/ext4-ialloc-uid-gid-and-pass-owner-down.patch base/ext4-projid-xattrs.patch linux-5.8/ext4-enc-flag.patch base/ext4-delayed-iput.patch +linux-5.10/ext4-fiemap-kernel-data.patch diff --git a/ldiskfs/kernel_patches/series/ldiskfs-5.14-rhel9.series b/ldiskfs/kernel_patches/series/ldiskfs-5.14-rhel9.series index 9da50d2..29c0d62 100644 --- a/ldiskfs/kernel_patches/series/ldiskfs-5.14-rhel9.series +++ b/ldiskfs/kernel_patches/series/ldiskfs-5.14-rhel9.series @@ -28,3 +28,4 @@ linux-5.14/ext4-ialloc-uid-gid-and-pass-owner-down.patch linux-5.14/ext4-projid-xattrs.patch base/ext4-delayed-iput.patch linux-5.14/ext4-xattr-disable-credits-check.patch +linux-5.10/ext4-fiemap-kernel-data.patch diff --git a/ldiskfs/kernel_patches/series/ldiskfs-5.14.21-sles15sp4.series b/ldiskfs/kernel_patches/series/ldiskfs-5.14.21-sles15sp4.series index ec43e14..02acf80 100644 --- a/ldiskfs/kernel_patches/series/ldiskfs-5.14.21-sles15sp4.series +++ b/ldiskfs/kernel_patches/series/ldiskfs-5.14.21-sles15sp4.series @@ -28,3 +28,4 @@ linux-5.14/ext4-ialloc-uid-gid-and-pass-owner-down.patch linux-5.14/ext4-projid-xattrs.patch base/ext4-delayed-iput.patch linux-5.14/ext4-xattr-disable-credits-check.patch +linux-5.10/ext4-fiemap-kernel-data.patch diff --git a/lustre/osd-ldiskfs/osd_io.c b/lustre/osd-ldiskfs/osd_io.c index 82651ef..4730f6a 100644 --- a/lustre/osd-ldiskfs/osd_io.c +++ b/lustre/osd-ldiskfs/osd_io.c @@ -1398,6 +1398,22 @@ struct osd_fextent { unsigned int mapped:1; }; +#ifdef KERNEL_DS +#define DECLARE_MM_SEGMENT_T(name) mm_segment_t name +#define access_set_kernel(saved_fs, fei) \ +do { \ + saved_fs = get_fs(); \ + set_fs(KERNEL_DS); \ +} while (0) +#define access_unset_kernel(saved_fs, fei) set_fs((saved_fs)) +#else +#define DECLARE_MM_SEGMENT_T(name) +#define access_set_kernel(saved_fs, fei) \ + (fei)->fi_flags |= LDISKFS_FIEMAP_FLAG_MEMCPY +#define access_unset_kernel(saved_fs, fei) \ + (fei)->fi_flags &= ~(LDISKFS_FIEMAP_FLAG_MEMCPY) +#endif /* KERNEL_DS */ + static int osd_is_mapped(struct dt_object *dt, __u64 offset, struct osd_fextent *cached_extent) { @@ -1407,6 +1423,7 @@ static int osd_is_mapped(struct dt_object *dt, __u64 offset, struct fiemap_extent_info fei = { 0 }; struct fiemap_extent fe = { 0 }; int rc; + DECLARE_MM_SEGMENT_T(saved_fs); if (block >= cached_extent->start && block < cached_extent->end) return cached_extent->mapped; @@ -1420,8 +1437,9 @@ static int osd_is_mapped(struct dt_object *dt, __u64 offset, fei.fi_extents_max = 1; fei.fi_extents_start = &fe; - + access_set_kernel(saved_fs, &fei); rc = inode->i_op->fiemap(inode, &fei, offset, FIEMAP_MAX_OFFSET-offset); + access_unset_kernel(saved_fs, &fei); if (rc != 0) return 0; @@ -2678,6 +2696,7 @@ static int osd_fiemap_get(const struct lu_env *env, struct dt_object *dt, struct inode *inode = osd_dt_obj(dt)->oo_inode; u64 len; int rc; + DECLARE_MM_SEGMENT_T(saved_fs); LASSERT(inode); if (inode->i_op->fiemap == NULL) @@ -2697,7 +2716,9 @@ static int osd_fiemap_get(const struct lu_env *env, struct dt_object *dt, if (fieinfo.fi_flags & FIEMAP_FLAG_SYNC) filemap_write_and_wait(inode->i_mapping); + access_set_kernel(saved_fs, &fieinfo); rc = inode->i_op->fiemap(inode, &fieinfo, fm->fm_start, len); + access_unset_kernel(saved_fs, &fieinfo); fm->fm_flags = fieinfo.fi_flags; fm->fm_mapped_extents = fieinfo.fi_extents_mapped; -- 1.8.3.1