LOOP_CONFIGURE is, as far as I understand it, supposed to be a way to
combine LOOP_SET_FD and LOOP_SET_STATUS64 into a single syscall. When
using LOOP_SET_FD+LOOP_SET_STATUS64, a single uevent would be sent for
each partition found on the loop device after the second ioctl(), but
when using LOOP_CONFIGURE, no such uevent was being sent.
In the old setup, uevents are disabled for LOOP_SET_FD, but not for
LOOP_SET_STATUS64. This makes sense, as it prevents uevents being
sent for a partially configured device during LOOP_SET_FD — they're
only sent at the end of LOOP_SET_STATUS64. But for LOOP_CONFIGURE,
uevents were disabled for the entire operation, so that final
notification was never issued. To fix this, I've moved the
loop_reread_partitions() call, which causes the uevents to be issued,
to after uevents are re-enabled, matching the behaviour of the
LOOP_SET_FD+LOOP_SET_STATUS64 combination.
I noticed this because Busybox's losetup program recently changed from
using LOOP_SET_FD+LOOP_SET_STATUS64 to LOOP_CONFIGURE, and this broke
my setup, for which I want a notification from the kernel any time a
new partition becomes available.
Signed-off-by: Alyssa Ross <hi@xxxxxxxxx>
Fixes: 3448914e8cc5 ("loop: Add LOOP_CONFIGURE ioctl")
---
v1: https://lore.kernel.org/linux-block/20230221222847.607096-1-hi@xxxxxxxxx/
v1 was an RFC, because I was looking for advice on how to handle
distinguishing between LOOP_SET_FD with non-zero max_part (in which
case partscan will be true, but a uevent should not be emitted), and
LOOP_CONFIGURE (where a uevent should be emitted). I didn't hear
anything, but I did some experimentation of my own, and adding a
partscan_uevent parameter to distinguish between LOOP_SET_FD and
LOOP_CONFIGURE feels like the least bad solution to me.
drivers/block/loop.c | 22 +++++++++++++++-------
1 file changed, 15 insertions(+), 7 deletions(-)
diff --git a/drivers/block/loop.c b/drivers/block/loop.c
index 839373451c2b..f00a0209b522 100644
--- a/drivers/block/loop.c
+++ b/drivers/block/loop.c
@@ -992,7 +992,8 @@ loop_set_status_from_info(struct loop_device *lo,
static int loop_configure(struct loop_device *lo, fmode_t mode,
struct block_device *bdev,
- const struct loop_config *config)
+ const struct loop_config *config,
+ bool partscan_uevent)
{
struct file *file = fget(config->fd);
struct inode *inode;
@@ -1110,15 +1111,21 @@ static int loop_configure(struct loop_device *lo, fmode_t mode,
clear_bit(GD_SUPPRESS_PART_SCAN, &lo->lo_disk->state);
loop_global_unlock(lo, is_loop);
- if (partscan)
- loop_reread_partitions(lo);
if (!(mode & FMODE_EXCL))
bd_abort_claiming(bdev, loop_configure);
+ /*
+ * Now that we are done, reread the partitions with uevent
+ * re-enabled if appropriate to let userspace know about the
+ * changes.
+ */
+ dev_set_uevent_suppress(disk_to_dev(lo->lo_disk), !partscan_uevent);
+ if (partscan)
+ loop_reread_partitions(lo);
+ dev_set_uevent_suppress(disk_to_dev(lo->lo_disk), 0);
+
error = 0;
done:
- /* enable and uncork uevent now that we are done */
- dev_set_uevent_suppress(disk_to_dev(lo->lo_disk), 0);
return error;
out_unlock:
@@ -1130,6 +1136,7 @@ static int loop_configure(struct loop_device *lo, fmode_t mode,
fput(file);
/* This is safe: open() is still holding a reference. */
module_put(THIS_MODULE);
+ dev_set_uevent_suppress(disk_to_dev(lo->lo_disk), 0);
goto done;
}
@@ -1547,7 +1554,7 @@ static int lo_ioctl(struct block_device *bdev, fmode_t mode,
memset(&config, 0, sizeof(config));
config.fd = arg;
- return loop_configure(lo, mode, bdev, &config);
+ return loop_configure(lo, mode, bdev, &config, false);
}
case LOOP_CONFIGURE: {
struct loop_config config;
@@ -1555,7 +1562,7 @@ static int lo_ioctl(struct block_device *bdev, fmode_t mode,
if (copy_from_user(&config, argp, sizeof(config)))
return -EFAULT;
- return loop_configure(lo, mode, bdev, &config);
+ return loop_configure(lo, mode, bdev, &config, true);
}
case LOOP_CHANGE_FD:
return loop_change_fd(lo, bdev, arg);
--
2.37.1
