Re: [LTP] [PATCH v1] ioctl_loop01.c: Use proper device for partitioning

[Wei Gao via ltp <ltp@lists.linux.it>]

On Mon, Sep 01, 2025 at 12:38:05PM +0200, Petr Vorel wrote:
> Hi Wei,
> 
> > This is same patch used on ioctl09,the page cache of loop0 can cache old
> > version of the partition table which is then used by the partitioning
> > code. Fix the problem by calling parted against the loop device directly.
> 
> > Link: https://lore.kernel.org/ltp/20250829141932.31997-1-jack@suse.cz/
> > Signed-off-by: Wei Gao <wegao@suse.com>
> > ---
> >  testcases/kernel/syscalls/ioctl/ioctl_loop01.c | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> > diff --git a/testcases/kernel/syscalls/ioctl/ioctl_loop01.c b/testcases/kernel/syscalls/ioctl/ioctl_loop01.c
> > index c9137bf1e..5ee7a474a 100644
> > --- a/testcases/kernel/syscalls/ioctl/ioctl_loop01.c
> > +++ b/testcases/kernel/syscalls/ioctl/ioctl_loop01.c
> > @@ -98,7 +98,7 @@ static void verify_ioctl_loop(void)
> >  static void setup(void)
> >  {
> >  	int ret;
> > -	const char *const cmd_parted[] = {"parted", "-s", "test.img", "mklabel", "msdos", "mkpart",
> > +	const char *const cmd_parted[] = {"parted", "-s", dev_path, "mklabel", "msdos", "mkpart",
> >  	                                  "primary", "ext4", "1M", "10M", NULL};
> 
> Indeed I was able to trigger the same error:
> 
> [ 2642.979234] ioctl_loop01
> [ 2642.997424] loop8: detected capacity change from 0 to 20480
> [ 2643.015094]  loop8: p1
> [ 2643.040903] blk_update_request: I/O error, dev loop8, sector 20352 op 0x0:(READ) flags 0x80700 phys_seg 1 prio class 0
> [ 2643.042105] blk_update_request: I/O error, dev loop8, sector 20352 op 0x0:(READ) flags 0x0 phys_seg 1 prio class 0
> [ 2643.043361] Buffer I/O error on dev loop8p1, logical block 2288, async page read
> [ 2643.044539] __loop_clr_fd: partition scan of loop8 failed (rc=-16)
> [ 2643.097517] blk_update_request: I/O error, dev loop8, sector 20352 op 0x0:(READ) flags 0x80700 phys_seg 1 prio class 0
> [ 2643.098175] blk_update_request: I/O error, dev loop8, sector 20352 op 0x0:(READ) flags 0x0 phys_seg 1 prio class 0
> [ 2643.098767] Buffer I/O error on dev loop8p1, logical block 2288, async page read
> [ 2643.109133] ioctl_ficlone02
> [ 2643.109824] loop8: detected capacity change from 0 to 614400
> [ 2643.175605] /dev/zero: Can't open blockdev
> [ 2643.307531] EXT4-fs (loop8): mounting ext2 file system using the ext4 subsystem
> [ 2643.312401] EXT4-fs (loop8): mounted filesystem without journal. Opts: (null). Quota mode: none.
> [ 2643.471527] EXT4-fs (loop8): mounting ext3 file system using the ext4 subsystem
> [ 2643.476826] EXT4-fs (loop8): mounted filesystem with ordered data mode. Opts: (null). Quota mode: none.
> [ 2643.644734] EXT4-fs (loop8): mounted filesystem with ordered data mode. Opts: (null). Quota mode: none.
> 
> But you change introduces one TCONF:
> 
> --- old	2025-09-01 05:05:50.401105483 -0400
> +++ new	2025-09-01 05:05:01.925105483 -0400
> @@ -5,6 +5,7 @@
>  tst_kconfig.c:678: TINFO: CONFIG_FAULT_INJECTION kernel option detected which might slow the execution
>  tst_test.c:1827: TINFO: Overall timeout per run is 0h 02m 04s
>  tst_device.c:99: TINFO: Found free device 6 '/dev/loop6'
> +ioctl_loop01.c:119: TCONF: parted exited with 1
>  tst_buffers.c:57: TINFO: Test is using guarded buffers
>  ioctl_loop01.c:84: TPASS: /sys/block/loop6/loop/partscan = 0
>  ioctl_loop01.c:85: TPASS: /sys/block/loop6/loop/autoclear = 0
> @@ -12,18 +13,16 @@
>  ioctl_loop01.c:56: TPASS: get expected lo_flag 12
>  ioctl_loop01.c:58: TPASS: /sys/block/loop6/loop/partscan = 1
>  ioctl_loop01.c:59: TPASS: /sys/block/loop6/loop/autoclear = 1
> -ioctl_loop01.c:68: TPASS: access /dev/loop6p1 succeeds
> -ioctl_loop01.c:74: TPASS: access /sys/block/loop6/loop6p1 succeeds
> +ioctl_loop01.c:62: TINFO: Current environment doesn't have parted disk, skip it
>  ioctl_loop01.c:90: TINFO: Test flag can be clear
>  ioctl_loop01.c:56: TPASS: get expected lo_flag 8
>  ioctl_loop01.c:58: TPASS: /sys/block/loop6/loop/partscan = 1
>  ioctl_loop01.c:59: TPASS: /sys/block/loop6/loop/autoclear = 0
> -ioctl_loop01.c:68: TPASS: access /dev/loop6p1 succeeds
> -ioctl_loop01.c:74: TPASS: access /sys/block/loop6/loop6p1 succeeds
> +ioctl_loop01.c:62: TINFO: Current environment doesn't have parted disk, skip it
>  
>  Summary:
> -passed   13
> +passed   9
>  failed   0
>  broken   0
> -skipped  0
> +skipped  1
>  warnings 0
> 
> That means your change effectively disable that code. => NACK
> 
> in ioctl09.c uses the first command with "test.img", then loop device on the
> second run.
> 
> Surprisingly trying to attach helps 'parted' to use it, although it produces a
> warning:
> 
> +++ testcases/kernel/syscalls/ioctl/ioctl_loop01.c
> @@ -106,6 +106,7 @@ static void setup(void)
>                 tst_brk(TBROK, "Failed to find free loop device");
> 
>         tst_fill_file("test.img", 0, 1024 * 1024, 10);
> +       tst_attach_device(dev_path, "test.img");
> 
>         ret = tst_cmd(cmd_parted, NULL, NULL, TST_CMD_PASS_RETVAL);
>         switch (ret) {
> 
> tst_device.c:99: TINFO: Found free device 9 '/dev/loop9'
> tst_buffers.c:57: TINFO: Test is using guarded buffers
> tst_device.c:176: TWARN: ioctl(/dev/loop9, LOOP_SET_FD, test.img) failed: EBUSY (16)
Thanks for your investigation.
You add tst_attach_device in setup, then second call tst_attach_device
in verify_ioctl_loop trigger this TWARN, I will send second patch.
> ioctl_loop01.c:84: TPASS: /sys/block/loop9/loop/partscan = 0
> ioctl_loop01.c:85: TPASS: /sys/block/loop9/loop/autoclear = 0
> ioctl_loop01.c:86: TPASS: /sys/block/loop9/loop/backing_file = '/tmp/LTP_iocfMLSpX/test.img'
> ioctl_loop01.c:56: TPASS: get expected lo_flag 12
> ioctl_loop01.c:58: TPASS: /sys/block/loop9/loop/partscan = 1
> ioctl_loop01.c:59: TPASS: /sys/block/loop9/loop/autoclear = 1
> ioctl_loop01.c:68: TPASS: access /dev/loop9p1 succeeds
> ioctl_loop01.c:74: TPASS: access /sys/block/loop9/loop9p1 succeeds
> ioctl_loop01.c:90: TINFO: Test flag can be clear
> ioctl_loop01.c:56: TPASS: get expected lo_flag 8
> ioctl_loop01.c:58: TPASS: /sys/block/loop9/loop/partscan = 1
> ioctl_loop01.c:59: TPASS: /sys/block/loop9/loop/autoclear = 0
> ioctl_loop01.c:68: TPASS: access /dev/loop9p1 succeeds
> ioctl_loop01.c:74: TPASS: access /sys/block/loop9/loop9p1 succeeds
> 
> Of course TWARN is also no-go.
> @Jan @Cyril @Li Any hint what am I missing.
> 
> Besides missing there is missing TST_CMD_TCONF_ON_MISSING on the flags (see
> today's fix [1]), but that's minor.
> 
> Kind regards,
> Petr
> 
> [1] https://github.com/linux-test-project/ltp/commit/9691c4b2bea4f772d61ca9e9a93d2087c88f6040

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp