On 08.06.21 21:14, Vladimir Sementsov-Ogievskiy wrote:
08.06.2021 16:16, Paolo Bonzini wrote:
Even though it was only called for devices that have bs->sg set (which
must be character devices), sg_get_max_segments looked at /sys/dev/block
which only works for block devices.
On Linux the sg driver has its own way to provide the maximum number of
iovecs in a scatter/gather list, so add support for it. The block
device
path is kept because it will be reinstated in the next patches.
Signed-off-by: Paolo Bonzini <pbonz...@redhat.com>
---
block/file-posix.c | 11 +++++++++++
1 file changed, 11 insertions(+)
diff --git a/block/file-posix.c b/block/file-posix.c
index f37dfc10b3..536998a1d6 100644
--- a/block/file-posix.c
+++ b/block/file-posix.c
@@ -1180,6 +1180,17 @@ static int sg_get_max_segments(int fd)
goto out;
}
+ if (S_ISCHR(st.st_mode)) {
Why not check "if (bs->sg) {" instead? It seems to be more consistent
with issuing SG_ ioctl. Or what I miss?
I dismissed this in v3, because I didn’t understand why you’d raise this
point. The function is called sg_*(), and it’s only called if bs->sg is
true anyway. So clearly we can use SG_ ioctls, because the whole
function is intended only for SG devices anyway.
This time, I looked forward, and perhaps starting at patch 4 I can
understand where you’re coming from, because then the function is used
for host devices in general.
So now I don’t particularly mind. I think it’s still clear that if
there’s a host device here that’s a character device, then that’s going
to be an SG device, so I don’t really have a preference between
S_ISCHR() and bs->sg.
Max
+ if (ioctl(fd, SG_GET_SG_TABLESIZE, &ret) == 0) {
+ return ret;
+ }
+ return -ENOTSUP;
+ }
+
+ if (!S_ISBLK(st.st_mode)) {
+ return -ENOTSUP;
+ }
+
sysfspath =
g_strdup_printf("/sys/dev/block/%u:%u/queue/max_segments",
major(st.st_rdev), minor(st.st_rdev));
sysfd = open(sysfspath, O_RDONLY);