Skip to contentSkip to conversation

Add qerror message if the 'change' target filename can't be opened

PATCH In review Ryan Harper 4 participants 10 messages Mar 25, 2010 – Apr 1, 2010

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 Add qerror message if the 'change' target filename can't be opened

Currently when using the change command to switch the file in the cd drive the command doesn't complain if the file doesn't exit or can't be opened and the drive keeps the existing image. This patch adds a qerror_report call to print a message out indicating the failure. This error message can be used to catch failures.

Current behavior:

QEMU 0.12.50 monitor - type 'help' for more information
(qemu) info block
ide0-hd0: type=hd removable=0 file=/dev/null ro=0 drv=host_device encrypted=0
ide1-cd0: type=cdrom removable=1 locked=0 [not inserted]
floppy0: type=floppy removable=1 locked=0 [not inserted]
sd0: type=floppy removable=1 locked=0 [not inserted]
(qemu) change ide1-cd0 /home/rharper/work/isos/Fedora-9-i386-DVD.iso (qemu) info block
ide0-hd0: type=hd removable=0 file=/dev/null ro=0 drv=host_device encrypted=0
ide1-cd0: type=cdrom removable=1 locked=0 file=/home/rharper/work/isos/Fedora-9-i386-DVD.iso ro=0 drv=raw encrypted=0
floppy0: type=floppy removable=1 locked=0 [not inserted]
sd0: type=floppy removable=1 locked=0 [not inserted]
(qemu) change ide1-cd0 /tmp/non_existent_file.iso
(qemu) info block
ide0-hd0: type=hd removable=0 file=/dev/null ro=0 drv=host_device encrypted=0
ide1-cd0: type=cdrom removable=1 locked=0 [not inserted]
floppy0: type=floppy removable=1 locked=0 [not inserted]
sd0: type=floppy removable=1 locked=0 [not inserted]
(qemu)

With patch:
QEMU 0.12.50 monitor - type 'help' for more information (qemu) change ide1-cd0 /tmp/non_existent_file.iso Could not open '/tmp/non_existent_file.iso' (qemu)

---
 monitor.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)
monitor.c +1 -0
index 0448a70..196c7a6 100644
@@ -1099,6 +1099,7 @@ static int do_change_block(Monitor *mon, const char *device,
         return -1;
     }
     if (bdrv_open2(bs, filename, BDRV_O_RDWR, drv) < 0) {
+        qerror_report(QERR_OPEN_FILE_FAILED, filename);
         return -1;
     }
     return monitor_read_bdrv_key_start(mon, bs, NULL, NULL);
Signature · 8 lines
1.6.3.3


-- 
Ryan Harper
Software Engineer; Linux Technology Center
IBM Corp., Austin, Tx
[email protected]
#1 lore raw
Quoted 6 lines · Ryan Harper

Currently when using the change command to switch the file in the cd drive the command doesn't complain if the file doesn't exit or can't be opened and the drive keeps the existing image. This patch adds a qerror_report call to print a message out indicating the failure. This error message can be used to catch failures.

[...]

Quoted 12 lines
---
 monitor.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)
monitor.c +0 -0
index 0448a70..196c7a6 100644

monitor.c static int do_change_block(Monitor *mon, const char *device, · line 1105

Quoted 7 lines
monitor.c +1 -0
@@ -1099,6 +1099,7 @@ static int do_change_block(Monitor *mon, const char *device,
         return -1;
     }
     if (bdrv_open2(bs, filename, BDRV_O_RDWR, drv) < 0) {
+        qerror_report(QERR_OPEN_FILE_FAILED, filename);
         return -1;
     }
     return monitor_read_bdrv_key_start(mon, bs, NULL, NULL);

We want this fix for QMP. Without it, we get UndefinedError, and a complaint ifdef CONFIG_DEBUG_MONITOR.

#2 lore raw
Quoted 15 lines · Markus Armbruster
Quoted 6 lines · Ryan Harper

Currently when using the change command to switch the file in the cd drive the command doesn't complain if the file doesn't exit or can't be opened and the drive keeps the existing image. This patch adds a qerror_report call to print a message out indicating the failure. This error message can be used to catch failures.

[...]

Quoted 6 lines
---
 monitor.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)
Quoted 16 lines

monitor.c static int do_change_block(Monitor *mon, const char *device, · line 1105

Quoted 13 lines
monitor.c +1 -0
index 0448a70..196c7a6 100644
@@ -1099,6 +1099,7 @@ static int do_change_block(Monitor *mon, const char *device,
         return -1;
     }
     if (bdrv_open2(bs, filename, BDRV_O_RDWR, drv) < 0) {
+        qerror_report(QERR_OPEN_FILE_FAILED, filename);
         return -1;
     }
     return monitor_read_bdrv_key_start(mon, bs, NULL, NULL);

We want this fix for QMP. Without it, we get UndefinedError, and a complaint ifdef CONFIG_DEBUG_MONITOR.

I'm not terribly familiar with the QMP stuff. Are you looking for me to fix some some stuff in the QMP bits? Are you just indicating that you need this fix for QMP?

Signature · 4 lines
Ryan Harper
Software Engineer; Linux Technology Center
IBM Corp., Austin, Tx
[email protected]
#3 lore raw
Quoted 33 lines · Ryan Harper
Quoted 15 lines · Markus Armbruster
Quoted 6 lines · Ryan Harper

Currently when using the change command to switch the file in the cd drive the command doesn't complain if the file doesn't exit or can't be opened and the drive keeps the existing image. This patch adds a qerror_report call to print a message out indicating the failure. This error message can be used to catch failures.

[...]

Quoted 6 lines
---
 monitor.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)
Quoted 16 lines

monitor.c static int do_change_block(Monitor *mon, const char *device, · line 1105

Quoted 13 lines
monitor.c +1 -0
index 0448a70..196c7a6 100644
@@ -1099,6 +1099,7 @@ static int do_change_block(Monitor *mon, const char *device,
         return -1;
     }
     if (bdrv_open2(bs, filename, BDRV_O_RDWR, drv) < 0) {
+        qerror_report(QERR_OPEN_FILE_FAILED, filename);
         return -1;
     }
     return monitor_read_bdrv_key_start(mon, bs, NULL, NULL);

We want this fix for QMP. Without it, we get UndefinedError, and a complaint ifdef CONFIG_DEBUG_MONITOR.

Quoted 3 lines

I'm not terribly familiar with the QMP stuff. Are you looking for me to fix some some stuff in the QMP bits? Are you just indicating that you need this fix for QMP?

The latter.

Thanks for fixing!

#4 lore raw
In reply to Ryan Harper
Quoted 5 lines · On Thu, 25 Mar 2010 09:32:58 -0500 Ryan Harper <[email protected]> wrote:

Currently when using the change command to switch the file in the cd drive the command doesn't complain if the file doesn't exit or can't be opened and the drive keeps the existing image. This patch adds a qerror_report call to print a message out indicating the failure. This error message can be used to catch failures.

Looks good to me, but it doesn't keep the existing image, it will silently
eject it instead.

Quoted 48 lines

Current behavior:

QEMU 0.12.50 monitor - type 'help' for more information
(qemu) info block
ide0-hd0: type=hd removable=0 file=/dev/null ro=0 drv=host_device encrypted=0
ide1-cd0: type=cdrom removable=1 locked=0 [not inserted]
floppy0: type=floppy removable=1 locked=0 [not inserted]
sd0: type=floppy removable=1 locked=0 [not inserted]
(qemu) change ide1-cd0 /home/rharper/work/isos/Fedora-9-i386-DVD.iso (qemu) info block
ide0-hd0: type=hd removable=0 file=/dev/null ro=0 drv=host_device encrypted=0
ide1-cd0: type=cdrom removable=1 locked=0 file=/home/rharper/work/isos/Fedora-9-i386-DVD.iso ro=0 drv=raw encrypted=0
floppy0: type=floppy removable=1 locked=0 [not inserted]
sd0: type=floppy removable=1 locked=0 [not inserted]
(qemu) change ide1-cd0 /tmp/non_existent_file.iso
(qemu) info block
ide0-hd0: type=hd removable=0 file=/dev/null ro=0 drv=host_device encrypted=0
ide1-cd0: type=cdrom removable=1 locked=0 [not inserted]
floppy0: type=floppy removable=1 locked=0 [not inserted]
sd0: type=floppy removable=1 locked=0 [not inserted]
(qemu)

With patch:
QEMU 0.12.50 monitor - type 'help' for more information (qemu) change ide1-cd0 /tmp/non_existent_file.iso Could not open '/tmp/non_existent_file.iso' (qemu)

---
 monitor.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)
monitor.c +1 -0
index 0448a70..196c7a6 100644
@@ -1099,6 +1099,7 @@ static int do_change_block(Monitor *mon, const char *device,
         return -1;
     }
     if (bdrv_open2(bs, filename, BDRV_O_RDWR, drv) < 0) {
+        qerror_report(QERR_OPEN_FILE_FAILED, filename);
         return -1;
     }
     return monitor_read_bdrv_key_start(mon, bs, NULL, NULL);
#5 lore raw
Quoted 11 lines · On Thu, 1 Apr 2010 15:15:51 -0300 Luiz Capitulino <[email protected]>…
Quoted 5 lines · On Thu, 25 Mar 2010 09:32:58 -0500 Ryan Harper <[email protected]> wrote:

Currently when using the change command to switch the file in the cd drive the command doesn't complain if the file doesn't exit or can't be opened and the drive keeps the existing image. This patch adds a qerror_report call to print a message out indicating the failure. This error message can be used to catch failures.

Looks good to me, but it doesn't keep the existing image, it will silently
eject it instead.

And, thinking more about it this seems the wrong behavior to me, if it
fails to open the file, it should not touch the current one.

Am I right, Kevin?

#6 lore raw

Am 01.04.2010 20:22, schrieb Luiz Capitulino:

Quoted 12 lines
Quoted 9 lines · On Thu, 1 Apr 2010 15:15:51 -0300 Luiz Capitulino <[email protected]>…
Quoted 5 lines · On Thu, 25 Mar 2010 09:32:58 -0500 Ryan Harper <[email protected]> wrote:

Currently when using the change command to switch the file in the cd drive the command doesn't complain if the file doesn't exit or can't be opened and the drive keeps the existing image. This patch adds a qerror_report call to print a message out indicating the failure. This error message can be used to catch failures.

Quoted 7 lines
Quoted 2 lines

Looks good to me, but it doesn't keep the existing image, it will silently
eject it instead.

And, thinking more about it this seems the wrong behavior to me, if it
fails to open the file, it should not touch the current one.

Am I right, Kevin?

Well, it's a monitor command. I guess its meaning has never been clearly defined, so I tend to say there is no right or wrong. From a user perspective, intuitively I would expect that it succeeds completely or maintains the old state, so yes.

However, I don't think it's fixable that easy. You obviously need to close the image before you can open it. And reopening the image in case of failure - well, we just had this discussion and I'm sure Juan wants to comment on it...

It's a case that we should consider if/when we reorganize bdrv_open. The "bdrv_close, but without closing the fd" thing doesn't work out here because we need to reuse the same bs. Or maybe open a different bs first and then copy things over. Could actually work this way.

Kevin

#7 lore raw
Quoted 19 lines · On Thu, 01 Apr 2010 21:14:30 +0200 Kevin Wolf <[email protected]> wrote:

Am 01.04.2010 20:22, schrieb Luiz Capitulino:

Quoted 18 lines
Quoted 11 lines · On Thu, 1 Apr 2010 15:15:51 -0300 Luiz Capitulino <[email protected]>…
Quoted 5 lines · On Thu, 25 Mar 2010 09:32:58 -0500 Ryan Harper <[email protected]> wrote:

Currently when using the change command to switch the file in the cd drive the command doesn't complain if the file doesn't exit or can't be opened and the drive keeps the existing image. This patch adds a qerror_report call to print a message out indicating the failure. This error message can be used to catch failures.

Looks good to me, but it doesn't keep the existing image, it will silently
eject it instead.

And, thinking more about it this seems the wrong behavior to me, if it
fails to open the file, it should not touch the current one.

Quoted 6 lines
Quoted 1 line

Am I right, Kevin?

Well, it's a monitor command. I guess its meaning has never been clearly defined, so I tend to say there is no right or wrong. From a user perspective, intuitively I would expect that it succeeds completely or maintains the old state, so yes.

Yes and I'd expect the same from a QMP perspective.

Quoted 9 lines

However, I don't think it's fixable that easy. You obviously need to close the image before you can open it. And reopening the image in case of failure - well, we just had this discussion and I'm sure Juan wants to comment on it...

It's a case that we should consider if/when we reorganize bdrv_open. The "bdrv_close, but without closing the fd" thing doesn't work out here because we need to reuse the same bs. Or maybe open a different bs first and then copy things over. Could actually work this way.

Yeah, I looked there and realized it wasn't easy but this solution
looks good.

#8 lore raw
In reply to Luiz Capitulino
Quoted 11 lines · Luiz Capitulino
Quoted 5 lines · On Thu, 25 Mar 2010 09:32:58 -0500 Ryan Harper <[email protected]> wrote:

Currently when using the change command to switch the file in the cd drive the command doesn't complain if the file doesn't exit or can't be opened and the drive keeps the existing image. This patch adds a qerror_report call to print a message out indicating the failure. This error message can be used to catch failures.

Looks good to me, but it doesn't keep the existing image, it will silently
eject it instead.

That's true, but that happens whether or not we indicate the change failed. I think fixing that is a good idea, but it shouldn't part of this patch (just indicating the failure).

Signature · 4 lines
Ryan Harper
Software Engineer; Linux Technology Center
IBM Corp., Austin, Tx
[email protected]
#9 lore raw
Quoted 10 lines · On Thu, 1 Apr 2010 13:23:44 -0500 Ryan Harper <[email protected]> wrote:
Quoted 9 lines · Luiz Capitulino
Quoted 5 lines · On Thu, 25 Mar 2010 09:32:58 -0500 Ryan Harper <[email protected]> wrote:

Currently when using the change command to switch the file in the cd drive the command doesn't complain if the file doesn't exit or can't be opened and the drive keeps the existing image. This patch adds a qerror_report call to print a message out indicating the failure. This error message can be used to catch failures.

Quoted 6 lines
Quoted 2 lines

Looks good to me, but it doesn't keep the existing image, it will silently
eject it instead.

That's true, but that happens whether or not we indicate the change failed. I think fixing that is a good idea, but it shouldn't part of this patch (just indicating the failure).

Sure, this patch is good by itself.