Re: [PATCH v2] x86: bring back rep movsq for user access on CPUs without ERMS

[Linus Torvalds]

On Sun, 3 Sept 2023 at 11:49, Linus Torvalds
<torvalds@xxxxxxxxxxxxxxxxxxxx> wrote:
>
> So I have no idea why you claim that "currently they have no choice".
> glibc is simply being incredibly stupid, and using newfstatat() for no
> good reason.

Do you have any good benchmark that shows the effects of this?

And if you do, does the attached patch (ENTIRELY UNTESTED!) fix the
silly glibc mis-feature?

               Linus
 fs/stat.c | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/fs/stat.c b/fs/stat.c
index f19b124974f2..e187dc79a313 100644
--- a/fs/stat.c
+++ b/fs/stat.c
@@ -311,6 +311,15 @@ int vfs_fstatat(int dfd, const char __user *filename,
 	int statx_flags = flags | AT_NO_AUTOMOUNT;
 	struct filename *name;
 
+	/* Work around glibc being silly */
+	if (dfd >= 0 && flags == AT_EMPTY_PATH) {
+		char c;
+		if (!get_user(c, filename)) {
+			if (!c)
+				return vfs_fstat(dfd, stat);
+		}
+	}
+
 	name = getname_flags(filename, getname_statx_lookup_flags(statx_flags), NULL);
 	ret = vfs_statx(dfd, name, statx_flags, stat, STATX_BASIC_STATS);
 	putname(name);