在 2025/8/22 11:06, 陈涛涛 Taotao Chen 写道:
From: Taotao Chen
Without O_LARGEFILE, file->f_op->write_iter calls
generic_write_check_limits(), which enforces a 2GB (MAX_NON_LFS) limit,
causing -EFBIG on large writes.
In shmem_pwrite(), this error is later masked as -EIO due to the error
ha
From: Taotao Chen
Without O_LARGEFILE, file->f_op->write_iter calls
generic_write_check_limits(), which enforces a 2GB (MAX_NON_LFS) limit,
causing -EFBIG on large writes.
In shmem_pwrite(), this error is later masked as -EIO due to the error
handling order, leading to igt failure
From: Taotao Chen
shmem_pwrite() currently checks for short writes before negative error
codes, which can overwrite real errors (e.g., -EFBIG) with -EIO.
Reorder the checks to return negative errors first, then handle short
writes.
Signed-off-by: Taotao Chen
Reviewed-by: Andi Shyti
---
v2
From: Taotao Chen
Change the address_space_operations callbacks write_begin() and
write_end() to take struct kiocb * as the first argument instead of
struct file *.
Update all affected function prototypes, implementations, call sites,
and related documentation across VFS, filesystems, and block
From: Taotao Chen
Set FOP_DONTCACHE in ext4_file_operations to declare support for
uncached buffered I/O.
To handle this flag, update ext4_write_begin() and ext4_da_write_begin()
to use write_begin_get_folio(), which encapsulates FGP_DONTCACHE logic
based on iocb->ki_flags.
Part of a ser
From: Taotao Chen
Refactors shmem_pwrite() to replace the ->write_begin/end logic
with a write_iter-based implementation using kiocb and iov_iter.
While kernel_write() was considered, it caused about 50% performance
regression. vfs_write() is not exported for kernel use. Therefore,
file-&g
From: Taotao Chen
Replace the write_begin/write_end loop in
i915_gem_object_create_shmem_from_data() with call to kernel_write().
This function initializes shmem-backed GEM objects. kernel_write()
simplifies the code by removing manual folio handling.
Part of a series refactoring
From: Taotao Chen
Add write_begin_get_folio() to simplify the common folio lookup logic
used by filesystem ->write_begin() implementations.
This helper wraps __filemap_get_folio() with common flags such as
FGP_WRITEBEGIN, conditional FGP_DONTCACHE, and set folio order based
on the write len
From: Taotao Chen
This patch series refactors the address_space_operations write_begin()
and write_end() callbacks to take const struct kiocb * as their first
argument, allowing IOCB flags such as IOCB_DONTCACHE to propagate to the
filesystem's buffered I/O path.
Ext4 is updated to impl
在 2025/7/14 17:11, Christian Brauner 写道:
On Thu, 10 Jul 2025 10:14:06 +, 陈涛涛 Taotao Chen wrote:
From: Taotao Chen
This patch series refactors the address_space_operations write_begin()
and write_end() callbacks to take const struct kiocb * as their first
argument, allowing IOCB flags
From: Taotao Chen
Change the address_space_operations callbacks write_begin() and
write_end() to take struct kiocb * as the first argument instead of
struct file *.
Update all affected function prototypes, implementations, call sites,
and related documentation across VFS, filesystems, and block
From: Taotao Chen
Add write_begin_get_folio() to simplify the common folio lookup logic
used by filesystem ->write_begin() implementations.
This helper wraps __filemap_get_folio() with common flags such as
FGP_WRITEBEGIN, conditional FGP_DONTCACHE, and set folio order based
on the write len
From: Taotao Chen
Refactors shmem_pwrite() to replace the ->write_begin/end logic
with a write_iter-based implementation using kiocb and iov_iter.
While kernel_write() was considered, it caused about 50% performance
regression. vfs_write() is not exported for kernel use. Therefore,
file-&g
From: Taotao Chen
This patch series refactors the address_space_operations write_begin()
and write_end() callbacks to take const struct kiocb * as their first
argument, allowing IOCB flags such as IOCB_DONTCACHE to propagate to the
filesystem's buffered I/O path.
Ext4 is updated to impl
From: Taotao Chen
Replace the write_begin/write_end loop in
i915_gem_object_create_shmem_from_data() with call to kernel_write().
This function initializes shmem-backed GEM objects. kernel_write()
simplifies the code by removing manual folio handling.
Part of a series refactoring
From: Taotao Chen
Set FOP_DONTCACHE in ext4_file_operations to declare support for
uncached buffered I/O.
To handle this flag, update ext4_write_begin() and ext4_da_write_begin()
to use write_begin_get_folio(), which encapsulates FGP_DONTCACHE logic
based on iocb->ki_flags.
Part of a ser
在 2025/7/7 22:54, Matthew Wilcox 写道:
On Mon, Jul 07, 2025 at 07:00:33AM +, 陈涛涛 Taotao Chen wrote:
+++ b/mm/filemap.c
I think this should be a static inline function. I don't think it's
worth moving out of line. Of course if you have measurements that show
differently, you can
From: Taotao Chen
Set FOP_DONTCACHE in ext4_file_operations to declare support for
uncached buffered I/O.
To handle this flag, update ext4_write_begin() and ext4_da_write_begin()
to use write_begin_get_folio(), which encapsulates FGP_DONTCACHE logic
based on iocb->ki_flags.
Part of a ser
From: Taotao Chen
Add write_begin_get_folio() to simplify the common folio lookup logic
used by filesystem ->write_begin() implementations.
This helper wraps __filemap_get_folio() with common flags such as
FGP_WRITEBEGIN, conditional FGP_DONTCACHE, and set folio order based
on the write len
From: Taotao Chen
Change the address_space_operations callbacks write_begin() and
write_end() to take struct kiocb * as the first argument instead of
struct file *.
Update all affected function prototypes, implementations, call sites,
and related documentation across VFS, filesystems, and block
From: Taotao Chen
This patch series refactors the address_space_operations write_begin()
and write_end() callbacks to take const struct kiocb * as their first
argument, allowing IOCB flags such as IOCB_DONTCACHE to propagate to the
filesystem's buffered I/O path.
Ext4 is updated to impl
From: Taotao Chen
Replace the write_begin/write_end loop in
i915_gem_object_create_shmem_from_data() with call to kernel_write().
This function initializes shmem-backed GEM objects. kernel_write()
simplifies the code by removing manual folio handling.
Part of a series refactoring
From: Taotao Chen
Refactors shmem_pwrite() to replace the ->write_begin/end logic
with a write_iter-based implementation using kiocb and iov_iter.
While kernel_write() was considered, it caused about 50% performance
regression. vfs_write() is not exported for kernel use. Therefore,
file-&g
在 2025/6/27 23:45, Matthew Wilcox 写道:
On Fri, Jun 27, 2025 at 11:03:11AM +, 陈涛涛 Taotao Chen wrote:
diff --git a/fs/exfat/file.c b/fs/exfat/file.c
index 841a5b18e3df..fdc2fa1e5c41 100644
--- a/fs/exfat/file.c
+++ b/fs/exfat/file.c
@@ -532,10 +532,12 @@ int exfat_file_fsync(struct file
在 2025/6/27 23:52, Matthew Wilcox 写道:
On Fri, Jun 27, 2025 at 11:03:11AM +, 陈涛涛 Taotao Chen wrote:
@@ -1399,13 +1400,10 @@ static int write_end_fn(handle_t *handle, struct inode
*inode,
}
/*
- * We need to pick up the new inode size which generic_commit_write gave us
- * `file
From: Taotao Chen
Set FOP_DONTCACHE in ext4_file_operations to declare support for
uncached buffered I/O.
To handle this flag, add processing for IOCB_DONTCACHE in
ext4_write_begin() and ext4_da_write_begin() by passing FGP_DONTCACHE
to page cache lookups.
Part of a series refactoring
From: Taotao Chen
Refactors shmem_pwrite() to replace the ->write_begin/end logic
with a write_iter-based implementation using kiocb and iov_iter.
While kernel_write() was considered, it caused about 50% performance
regression. vfs_write() is not exported for kernel use. Therefore,
file-&g
From: Taotao Chen
Change the address_space_operations callbacks write_begin() and
write_end() to take struct kiocb * as the first argument instead of
struct file *.
Update all affected function prototypes, implementations, call sites,
and related documentation across VFS, filesystems, and block
From: Taotao Chen
Replace the write_begin/write_end loop in
i915_gem_object_create_shmem_from_data() with call to kernel_write().
This function initializes shmem-backed GEM objects. kernel_write()
simplifies the code by removing manual folio handling.
Part of a series refactoring
From: Taotao Chen
This patch series refactors the address_space_operations write_begin()
and write_end() callbacks to take const struct kiocb * as their first
argument, allowing IOCB flags such as IOCB_DONTCACHE to propagate to the
filesystem's buffered I/O path.
Ext4 is updated to impl
From: Taotao Chen
Refactors shmem_pwrite() to replace the ->write_begin/end logic
with a write_iter-based implementation using kiocb and iov_iter.
While kernel_write() was considered, it caused about 50% performance
regression. vfs_write() is not exported for kernel use. Therefore,
file-&g
From: Taotao Chen
Change the address_space_operations callbacks write_begin() and
write_end() to take struct kiocb * as the first argument instead of
struct file *.
Update all affected function prototypes, implementations, call sites,
and related documentation across VFS, filesystems, and block
From: Taotao Chen
Set the FOP_DONTCACHE flag in ext4_file_operations to indicate that
ext4 supports IOCB_DONTCACHE handling in buffered write paths.
Part of a series refactoring address_space_operations write_begin and
write_end callbacks to use struct kiocb for passing write context and
flags
From: Taotao Chen
Replace the write_begin/write_end loop in
i915_gem_object_create_shmem_from_data() with call to kernel_write().
This function initializes shmem-backed GEM objects. kernel_write()
simplifies the code by removing manual folio handling.
Part of a series refactoring
From: Taotao Chen
This patch series refactors the address_space_operations write_begin()
and write_end() callbacks to take struct kiocb * as their first argument,
allowing IOCB flags such as IOCB_DONTCACHE to propagate to filesystem’s
buffered write path.
Ext4 is updated to implement handling
From: Taotao Chen
Add support for the IOCB_DONTCACHE flag in ext4_write_begin() and
ext4_da_write_begin(). When set in the kiocb, the FGP_DONTCACHE bit
is passed to the page cache lookup, preventing written pages from
being retained in the cache.
Only the handling logic is implemented here; the
36 matches
Mail list logo
