On Wed, 22 Mar 2023 at 12:14, Jonathan Wakely wrote:
>
>
> On Mon, 20 Mar 2023 at 22:30, Jonathan Wakely via Libstdc++ <
> libstd...@gcc.gnu.org> wrote:
>
>> On 20/03/23 22:27 +0000, Jonathan Wakely wrote:
>> >On 06/03/23 20:52 +0100, Jannik Glückert wrote:
>> >>we were previously only using sendfile for files smaller than 2GB, as
>> >>sendfile needs to be called repeatedly for files bigger than that.
>> >>
>> >>some quick numbers, copying a 16GB file, average of 10 repetitions:
>> >> old:
>> >> real: 13.4s
>> >> user: 0.14s
>> >> sys : 7.43s
>> >> new:
>> >> real: 8.90s
>> >> user: 0.00s
>> >> sys : 3.68s
>> >>
>> >>Additionally, this fixes
>> >>https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108178
>> >>
>> >>libstdc++-v3/ChangeLog:
>> >>
>> >> * acinclude.m4 (_GLIBCXX_HAVE_LSEEK): define
>> >> * config.h.in: Regenerate.
>> >> * configure: Regenerate.
>> >> * src/filesystem/ops-common.h: enable sendfile for files
>> >> >2GB in std::filesystem::copy_file, skip zero-length files
>>
>> Also, the ChangeLog entry needs to be indented with tabs, name the
>> changed functions, and should be complete sentences, e.g.
>>
>> * acinclude.m4 (_GLIBCXX_HAVE_LSEEK): Define.
>> * config.h.in: Regenerate.
>> * configure: Regenerate.
>> * src/filesystem/ops-common.h (copy_file_sendfile): Define new
>> function for sendfile logic. Loop to support large files. Skip
>> zero-length files.
>> (do_copy_file): Use it.
>>
>>
> Here's what I plan to commit in a few weeks when GCC 14 Stage 1 opens.
>
>
>
And similarly for the copy_file_range change.
commit 2ad500e358c03ef63af1540d44645df582a4809c
Author: Jannik Glückert <jannik.glueck...@gmail.com>
Date: Wed Mar 8 18:37:43 2023
libstdc++: Use copy_file_range for filesystem::copy_file
copy_file_range is a recent-ish syscall for copying files. It is similar
to sendfile but allows filesystem-specific optimizations. Common are:
Reflinks: BTRFS, XFS, ZFS (does not implement the syscall yet)
Server-side copy: NFS, SMB, Ceph
If copy_file_range is not available for the given files, fall back to
sendfile / userspace copy.
libstdc++-v3/ChangeLog:
* acinclude.m4 (_GLIBCXX_USE_COPY_FILE_RANGE): Define.
* config.h.in: Regenerate.
* configure: Regenerate.
* src/filesystem/ops-common.h (copy_file_copy_file_range):
Define new function.
(do_copy_file): Use it.
Signed-off-by: Jannik Glückert <jannik.glueck...@gmail.com>
diff --git a/libstdc++-v3/acinclude.m4 b/libstdc++-v3/acinclude.m4
index 85a09a5a869..4cf02dc6e4e 100644
--- a/libstdc++-v3/acinclude.m4
+++ b/libstdc++-v3/acinclude.m4
@@ -4581,6 +4581,7 @@ dnl _GLIBCXX_USE_UTIMENSAT
dnl _GLIBCXX_USE_ST_MTIM
dnl _GLIBCXX_USE_FCHMOD
dnl _GLIBCXX_USE_FCHMODAT
+dnl _GLIBCXX_USE_COPY_FILE_RANGE
dnl _GLIBCXX_USE_SENDFILE
dnl HAVE_LINK
dnl HAVE_LSEEK
@@ -4779,6 +4780,25 @@ dnl
if test $glibcxx_cv_truncate = yes; then
AC_DEFINE(HAVE_TRUNCATE, 1, [Define if truncate is available in
<unistd.h>.])
fi
+dnl
+ AC_CACHE_CHECK([for copy_file_range that can copy files],
+ glibcxx_cv_copy_file_range, [dnl
+ case "${target_os}" in
+ linux*)
+ GCC_TRY_COMPILE_OR_LINK(
+ [#include <unistd.h>],
+ [copy_file_range(1, nullptr, 2, nullptr, 1, 0);],
+ [glibcxx_cv_copy_file_range=yes],
+ [glibcxx_cv_copy_file_range=no])
+ ;;
+ *)
+ glibcxx_cv_copy_file_range=no
+ ;;
+ esac
+ ])
+ if test $glibcxx_cv_copy_file_range = yes; then
+ AC_DEFINE(_GLIBCXX_USE_COPY_FILE_RANGE, 1, [Define if copy_file_range is
available in <unistd.h>.])
+ fi
dnl
AC_CACHE_CHECK([for sendfile that can copy files],
glibcxx_cv_sendfile, [dnl
diff --git a/libstdc++-v3/src/filesystem/ops-common.h
b/libstdc++-v3/src/filesystem/ops-common.h
index 7874a95488a..906436b484e 100644
--- a/libstdc++-v3/src/filesystem/ops-common.h
+++ b/libstdc++-v3/src/filesystem/ops-common.h
@@ -49,6 +49,9 @@
#ifdef NEED_DO_COPY_FILE
# include <filesystem>
# include <ext/stdio_filebuf.h>
+# ifdef _GLIBCXX_USE_COPY_FILE_RANGE
+# include <unistd.h> // copy_file_range
+# endif
# ifdef _GLIBCXX_USE_SENDFILE
# include <sys/sendfile.h> // sendfile
# include <unistd.h> // lseek
@@ -359,6 +362,32 @@ _GLIBCXX_BEGIN_NAMESPACE_FILESYSTEM
}
#ifdef NEED_DO_COPY_FILE
+#ifdef _GLIBCXX_USE_COPY_FILE_RANGE
+ bool
+ copy_file_copy_file_range(int fd_in, int fd_out, size_t length) noexcept
+ {
+ // a zero-length file is either empty, or not copyable by this syscall
+ // return early to avoid the syscall cost
+ if (length == 0)
+ {
+ errno = EINVAL;
+ return false;
+ }
+ size_t bytes_left = length;
+ off64_t off_in = 0, off_out = 0;
+ ssize_t bytes_copied;
+ do
+ {
+ bytes_copied = ::copy_file_range(fd_in, &off_in, fd_out, &off_out,
+ bytes_left, 0);
+ bytes_left -= bytes_copied;
+ }
+ while (bytes_left > 0 && bytes_copied > 0);
+ if (bytes_copied < 0)
+ return false;
+ return true;
+ }
+#endif
#if defined _GLIBCXX_USE_SENDFILE && ! defined _GLIBCXX_FILESYSTEM_IS_WINDOWS
bool
copy_file_sendfile(int fd_in, int fd_out, size_t length) noexcept
@@ -529,6 +558,33 @@ _GLIBCXX_BEGIN_NAMESPACE_FILESYSTEM
bool has_copied = false;
+#ifdef _GLIBCXX_USE_COPY_FILE_RANGE
+ if (!has_copied)
+ has_copied = copy_file_copy_file_range(in.fd, out.fd, from_st->st_size);
+ if (!has_copied)
+ {
+ // EINVAL: src and dst are the same file (this is not cheaply
+ // detectable from userspace)
+ // EINVAL: copy_file_range is unsupported for this file type by the
+ // underlying filesystem
+ // ENOTSUP: undocumented, can arise with old kernels and NFS
+ // EOPNOTSUPP: filesystem does not implement copy_file_range
+ // ETXTBSY: src or dst is an active swapfile (nonsensical, but allowed
+ // with normal copying)
+ // EXDEV: src and dst are on different filesystems that do not support
+ // cross-fs copy_file_range
+ // ENOENT: undocumented, can arise with CIFS
+ // ENOSYS: unsupported by kernel or blocked by seccomp
+ if (errno != EINVAL && errno != ENOTSUP && errno != EOPNOTSUPP
+ && errno != ETXTBSY && errno != EXDEV && errno != ENOENT
+ && errno != ENOSYS)
+ {
+ ec.assign(errno, std::generic_category());
+ return false;
+ }
+ }
+#endif
+
#if defined _GLIBCXX_USE_SENDFILE && ! defined _GLIBCXX_FILESYSTEM_IS_WINDOWS
if (!has_copied)
has_copied = copy_file_sendfile(in.fd, out.fd, from_st->st_size);
