fix FIO_OPT_STR_SET and update io_uring doc
A message that is not in the archive (sent off-list or not yet received): CGME20221021114404epcas5p279068edb23eb2a58902424ed123df48d@epcas5p2.samsung.com
cover fix FIO_OPT_STR_SET and update io_uring doc
This series has 2 small patches.
First patch deals with type FIO_OPT_STR_SET.
FIO_OPT_STR_SET suggests the passed value should be empty or boolean value ([0|1]).
But for the given type it accepts any integer value as there is no limit for min and max values. Add a limit for that during the option initialization.
Note: With this the existing job files, test scripts which used to pass any integer value will not work.
Other way to fix this would be to update the typehelp for FIO_OPT_STR_SET as "empty or integer value (opt=100)"
Second patch sets the type for sqthread_poll to FIO_OPT_STR_SET instead of FIO_OPT_INT. This I believe is what was originally intended when the option was added.
This patch also adds the missing types for io_uring options in the documentation.
This now fixes the issue:
https://github.com/axboe/fio/issues/927
Ankit Kumar (2):
parse: set min and max limit for FIO_OPT_STR_SET
io_uring: update documentation and small fix for sqthread_poll
HOWTO.rst | 6 +++--- engines/io_uring.c | 2 +- fio.1 | 6 +++--- parse.c | 2 +- 4 files changed, 8 insertions(+), 8 deletions(-)
Signature · 1 line
2.17.1
PATCH 1/2 parse: set min and max limit for FIO_OPT_STR_SET
type FIO_OPT_STR_SET suggests the passed value should be empty or boolean value ([0|1]).
But the given type accepts any integer value as there is no limit for min and max values. Add a limit to that.
--- parse.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
parse.c +1 -1
@@ -1434,7 +1434,7 @@ static void option_init(struct fio_option *o)return ;if (o->name && !o->lname)log_err ("Option %s: missing long option name\n" , o->name); -if (o->type ==FIO_OPT_BOOL ) { +if (o->type ==FIO_OPT_BOOL || o->type ==FIO_OPT_STR_SET ) { o->minval =0 ; o->maxval =1 ; }
Signature · 1 line
2.17.1
PATCH 2/2 io_uring: update documentation and small fix for sqthread_poll
In reply to Ankit Kumartype for sqthread_poll should be FIO_OPT_STR_SET, as this only requires it to be set. This also fixes the issue:
https://github.com/axboe/fio/issues/927
Update the HOWTO and man page to specify the missing types for all the remaining io_uring options.
--- HOWTO.rst | 6 +++--- engines/io_uring.c | 2 +- fio.1 | 6 +++--- 3 files changed, 7 insertions(+), 7 deletions(-)
HOWTO.rst +3 -3
@@ -2274,7 +2274,7 @@ with the caveat that when used on the command line, they must come after the map and release for each IO. This is more efficient, and reduces the IO latency as well. -.. option:: nonvectored : [io_uring] [io_uring_cmd] +.. option:: nonvectored=int : [io_uring] [io_uring_cmd] With this option, fio will use non-vectored read/write commands, where address must contain the address directly. Default is -1.
@@ -2301,7 +2301,7 @@ with the caveat that when used on the command line, they must come after the This frees up cycles for fio, at the cost of using more CPU in the system. -.. option:: sqthread_poll_cpu : [io_uring] [io_uring_cmd] +.. option:: sqthread_poll_cpu=int : [io_uring] [io_uring_cmd] When :option:`sqthread_poll` is set, this option provides a way to define which CPU should be used for the polling thread.
@@ -2351,7 +2351,7 @@ with the caveat that when used on the command line, they must come after the When hipri is set this determines the probability of a pvsync2 I/O being high priority. The default is 100%. -.. option:: nowait : [pvsync2] [libaio] [io_uring] +.. option:: nowait=bool : [pvsync2] [libaio] [io_uring] [io_uring_cmd] By default if a request cannot be executed immediately (e.g. resource starvation, waiting on locks) it is queued and the initiating process will be blocked until
engines/io_uring.c +1 -1
@@ -226,7 +226,7 @@ static struct fio_option options[] = { { .name = "sqthread_poll", .lname = "Kernel SQ thread polling", - .type = FIO_OPT_INT, + .type = FIO_OPT_STR_SET, .off1 = offsetof(struct ioring_options, sqpoll_thread), .help = "Offload submission/completion to kernel thread", .category = FIO_OPT_C_ENGINE,
fio.1 +3 -3
@@ -2063,7 +2063,7 @@ release them when IO is done. If this option is set, the pages are pre-mapped before IO is started. This eliminates the need to map and release for each IO. This is more efficient, and reduces the IO latency as well. .TP -.BI (io_uring,io_uring_cmd)nonvectored +.BI (io_uring,io_uring_cmd)nonvectored \fR=\fPint With this option, fio will use non-vectored read/write commands, where address must contain the address directly. Default is -1. .TP
@@ -2092,7 +2092,7 @@ available items in the SQ ring. If this option is set, the act of submitting IO will be done by a polling thread in the kernel. This frees up cycles for fio, at the cost of using more CPU in the system. .TP -.BI (io_uring,io_uring_cmd)sqthread_poll_cpu +.BI (io_uring,io_uring_cmd)sqthread_poll_cpu \fR=\fPint When `sqthread_poll` is set, this option provides a way to define which CPU should be used for the polling thread. .TP
@@ -2115,7 +2115,7 @@ than normal. When hipri is set this determines the probability of a pvsync2 I/O being high priority. The default is 100%. .TP -.BI (pvsync2,libaio,io_uring)nowait +.BI (pvsync2,libaio,io_uring,io_uring_cmd)nowait \fR=\fPbool By default if a request cannot be executed immediately (e.g. resource starvation, waiting on locks) it is queued and the initiating process will be blocked until the required resource becomes free.
Signature · 1 line
2.17.1
Quoted 29 lines · Ankit Kumar
This series has 2 small patches.
First patch deals with type
FIO_OPT_STR_SET.
FIO_OPT_STR_SETsuggests the passed value should be empty or boolean value ([0|1]).
But for the given type it accepts any integer value as there is no limit for min and max values. Add a limit for that during the option initialization.Note: With this the existing job files, test scripts which used to pass any integer value will not work.
Other way to fix this would be to update the typehelp for
FIO_OPT_STR_SETas "empty or integer value (opt=100)"Second patch sets the type for
sqthread_polltoFIO_OPT_STR_SETinstead ofFIO_OPT_INT. This I believe is what was originally intended when the option was added.
This patch also adds the missing types forio_uringoptions in the documentation.This now fixes the issue:
https://github.com/axboe/fio/issues/927Ankit Kumar (2):
parse: set min and max limit forFIO_OPT_STR_SET
io_uring: update documentation and small fix forsqthread_poll
Quoted 6 lines
HOWTO.rst | 6 +++--- engines/io_uring.c | 2 +- fio.1 | 6 +++--- parse.c | 2 +- 4 files changed, 8 insertions(+), 8 deletions(-)
Ankit, I merged the second patch that cleans up the documentation. The first patch may break existing job files and I'm not convinced that doing so is worth it, although I agree that your fix is what we should have done from the beginning.
Vincent