[osv-dev] [PATCH] headers: get rid of fenv.h copy

Signed-off-by: Waldemar Kozaczuk --- arch/aarch64/feexcept.cc | 1 + arch/x64/feexcept.cc | 1 + include/api/__fenv.h | 15 +++ include/api/fenv.h | 32 +--- tests/tst-feexcept.cc| 1 + 5 files changed, 19 insertions(+), 31

[osv-dev] [PATCH] libc: drop 5 more headers with symlinks to musl copies

Those 5 headers are mostly identical to their musl copies except they were modified to comment out *64 macros that serve as aliasing mechanism on musl side but conflicts in all places where we use LFS64 macro that uses weak_alias to provide symbol copies. So in order to avoid this collisions and

[osv-dev] [PATCH] libc: replace include/api/sys/eventfd.h with symlink to musl copy

The eventfd.h only differs by presense of declaration eventfd2() function which is not a Linux ABI function but syscall name and was probably added to the header to accomodate implementing eventfd2 as a syscall in linux.cc. So let us move that declaration to linux.cc where it is used.

[osv-dev] [PATCH] libc: replace include/api/aarch64/bits/mman.h with a symlink to musl copy

This file and its aarch64 equivalent differs only to musl copies by extra MAP_UNINITIALIZED macro. In order to drop this header we create extra __mman.h headers and use them in libc/mman.cc. Signed-off-by: Waldemar Kozaczuk --- include/api/aarch64/bits/__mman.h | 13 +++

[osv-dev] [PATCH] libc: replace include/api/x64/bits/socket.h with a symlink to musl copy

This file is identical to musl copy except for added #ifndef OSV_BITS_SOCKET_H ... which may have been necessary at some point but it is not now. Signed-off-by: Waldemar Kozaczuk --- include/api/x64/bits/socket.h | 23 +-- 1 file changed, 1 insertion(+), 22 deletions(-)

[osv-dev] [PATCH] api: replace 3 more include/api headers with musl copies

This patch replaces following 3 headers with symlinks to musl copies: - include/api/stdint.h - include/api/stdio.h - include/api/x64/bits/stdint.h These files come from older version of musl and have not be been touched on OSv side. However they had changed in newer version (0.9.12) of musl in a

Re: [osv-dev] Re: [PATCH 2/5] vfs: homogenize mount_rofs_rootfs and mount_zfs_rootfs

Let us see what others think (Nadav?). In my humble opinion, making loader.cc backward compatible would not make it all that much more complicated (which is really changing the last "else" in that "if" to do what old logic did). It would simply act as a fallback if the user did not specify

[osv-dev] [COMMIT osv master] virtio-fs: use exclusive device ids

From: Fotis Xenakis Committer: Waldemar Kozaczuk Branch: master virtio-fs: use exclusive device ids Previously, the devfs entry for each virtio-fs device was derived from virtio::virtio_driver::_disk_idx, which is used by virtio-blk and virtio-scsi, which anyway share a devfs namespace. This

[osv-dev] [COMMIT osv master] vfs: homogenize mount_rofs_rootfs and mount_zfs_rootfs

From: Fotis Xenakis Committer: Waldemar Kozaczuk Branch: master vfs: homogenize mount_rofs_rootfs and mount_zfs_rootfs Signed-off-by: Fotis Xenakis Message-Id: --- diff --git a/fs/vfs/main.cc b/fs/vfs/main.cc --- a/fs/vfs/main.cc +++ b/fs/vfs/main.cc @@ -1593,7 +1593,7 @@ int faccessat(int

Re: [osv-dev] Re: [PATCH 2/5] vfs: homogenize mount_rofs_rootfs and mount_zfs_rootfs

Indeed, there was no way for this change to be 100% backwards compatible, except the one you just described. Breaking tests.py was admittedly an omission on my side to test more thoroughly. What you suggest (change the loader's behavior) should be the option with the least effort involved and