Linux block layer
 help / color / mirror / Atom feed
From: Shin'ichiro Kawasaki <shinichiro.kawasaki@wdc.com>
To: Sebastian Chlad <sebastianchlad@gmail.com>
Cc: linux-block@vger.kernel.org,  Sebastian Chlad <sebastian.chlad@suse.com>
Subject: Re: [PATCH 1/2 blktests] src/miniublk: switch to ioctl-encoded ublk commands
Date: Fri, 19 Jun 2026 12:26:11 +0900	[thread overview]
Message-ID: <ajSud4Y4PmCu2X_5@shinmob> (raw)
In-Reply-To: <20260617072516.6238-2-sebastian.chlad@suse.com>

Hi Sebastian,

Thanks for the patches. I agree that this direction is good: it's the better
shift away from the legacy interface.

One point I noticed is that src/miniublk.c can no longer be built with the
kernel headers of the LTS kernel version v6.1.y, probably (v5.15.y does not have
ublk and v6.6.y supports the new interface). This is a rather small window, and
may be acceptable but I wonder what you think about it

If we drop the miniublk build with v6.1.y kernel headers, it might be the better
to check before building miniublk. I quickly created a Makefile change [1] for
that purpose.

Also, please find a comment in line below.

On Jun 17, 2026 / 09:25, Sebastian Chlad wrote:
> Kernels built without CONFIG_BLKDEV_UBLK_LEGACY_OPCODES reject the
> legacy raw UBLK_CMD_* and UBLK_IO_* opcodes. Switch miniublk to use
> the ioctl-encoded UBLK_U_CMD_* and UBLK_U_IO_* variants defined in
> linux/ublk_cmd.h instead.
> 
> For IO commands, the ioctl-encoded opcode is used for submission while
> _IOC_NR() extracts the raw NR bits for build_user_data(), keeping the
> user_data tag encoding intact.
> 
> Signed-off-by: Sebastian Chlad <sebastian.chlad@suse.com>
> ---
>  src/miniublk.c | 30 +++++++++++++++---------------
>  1 file changed, 15 insertions(+), 15 deletions(-)
> 
> diff --git a/src/miniublk.c b/src/miniublk.c
> index f98f850..5a35ca7 100644
> --- a/src/miniublk.c
> +++ b/src/miniublk.c
[...]
> @@ -624,9 +624,9 @@ static int ublk_queue_io_cmd(struct ublk_queue *q,
>  		return 0;
>  
>  	if (io->flags & UBLKSRV_NEED_COMMIT_RQ_COMP)
> -		cmd_op = UBLK_IO_COMMIT_AND_FETCH_REQ;
> -	else if (io->flags & UBLKSRV_NEED_FETCH_RQ)
> -		cmd_op = UBLK_IO_FETCH_REQ;
> +		cmd_op = UBLK_U_IO_COMMIT_AND_FETCH_REQ;
> +	else
> +		cmd_op = UBLK_U_IO_FETCH_REQ;

The hunk above changes the "else if" part, is this intentional?


[1]

diff --git a/src/Makefile b/src/Makefile
index d8833bf..adfe3ef 100644
--- a/src/Makefile
+++ b/src/Makefile
@@ -8,6 +8,10 @@ HAVE_C_MACRO = $(shell if echo "$(H)include <$(1)>" |	\
 		$(CC) $(CFLAGS) -E - 2>&1 /dev/null | grep $(2) > /dev/null 2>&1; \
 		then echo 1;else echo 0; fi)
 
+HAVE_C_DEF = $(shell if echo -e "$(H)include <$(1)>\n#ifdef $(2)\nHAVE_$(2)\n#endif" | \
+		$(CC) $(CFLAGS) -E - 2>&1 /dev/null | grep HAVE_$(2) > /dev/null 2>&1; \
+		then echo 1;else echo 0; fi)
+
 C_TARGETS := \
 	dio-offsets \
 	loblksize \
@@ -27,6 +31,7 @@ C_UBLK_TARGETS := miniublk
 
 HAVE_LIBURING := $(call HAVE_C_MACRO,liburing.h,IORING_OP_URING_CMD)
 HAVE_UBLK_HEADER := $(call HAVE_C_HEADER,linux/ublk_cmd.h,1)
+HAVE_NEW_UBLK_INTF := $(call HAVE_C_DEF,linux/ublk_cmd.h,UBLK_U_CMD_START_DEV)
 
 CXX_TARGETS := \
 	discontiguous-io
@@ -37,8 +42,12 @@ SYZKALLER_TARGETS := \
 TARGETS := $(C_TARGETS) $(CXX_TARGETS) $(SYZKALLER_TARGETS)
 
 ifeq ($(HAVE_UBLK_HEADER), 1)
+ifeq ($(HAVE_NEW_UBLK_INTF), 1)
 C_URING_TARGETS += $(C_UBLK_TARGETS)
 else
+$(info Skip $(C_UBLK_TARGETS) build due to missing new ublk interface(v6.4+))
+endif
+else
 $(info Skip $(C_UBLK_TARGETS) build due to missing kernel header(v6.0+))
 endif
 


  reply	other threads:[~2026-06-19  3:26 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-06-17  7:25 [PATCH 0/2 blktests] Update the miniublk to use ioctl opcodes Sebastian Chlad
2026-06-17  7:25 ` [PATCH 1/2 blktests] src/miniublk: switch to ioctl-encoded ublk commands Sebastian Chlad
2026-06-19  3:26   ` Shin'ichiro Kawasaki [this message]
2026-06-17  7:25 ` [PATCH 2/2 blktests] src/miniublk: fall back to legacy opcodes on older kernels Sebastian Chlad

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=ajSud4Y4PmCu2X_5@shinmob \
    --to=shinichiro.kawasaki@wdc.com \
    --cc=linux-block@vger.kernel.org \
    --cc=sebastian.chlad@suse.com \
    --cc=sebastianchlad@gmail.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox