Skip to contentSkip to conversation

Monitor-related fixes

PATCH Applied Stratos Psomadakis 4 participants 17 messages Sep 11, 2014 – Sep 15, 2014

lore
Share & tools
lore
#0 lore raw

cover Monitor-related fixes

Hi,

the first patch fixes an issue with HMP monitors, which was exposed with v2.1.0 (commits cdaa86a and 812c10). The second one fixes a typo in a helper C program used in qemu-iotests.

We think that they should be cherry-picked for the next stable release.

Thanks,
Stratos

Stratos Psomadakis (2):
monitor: Reset HMP mon->rs on CHR_EVENT_CLOSED
iotests: Send the correct fd in socket_scm_helper

 monitor.c                              |    1 +
 tests/qemu-iotests/socket_scm_helper.c |    2 +-
 2 files changed, 2 insertions(+), 1 deletion(-)
Signature · 1 line
1.7.10.4
#1 lore raw

PATCH 1/2 monitor: Reset HMP mon->rs on CHR_EVENT_CLOSED

Commit cdaa86a54 ("Add G_IO_HUP handler for socket chardev") exposed a bug in the way the HMP monitor handles its input. When a client closes the connection to the monitor, tcp_chr_read() will catch the HUP 'signal' and call tcp_chr_disconnect() to close the server-side connection too. Due to the fact that monitor reads 1 byte at a time (for each tcp_chr_read()), the monitor readline state / buffers can be left in an inconsistent state (i.e. a half-finished command). Thus, without calling readline_restart() on mon->rs upon CHR_EVENT_CLOSED, future HMP commands will fail.

---
 monitor.c |    1 +
 1 file changed, 1 insertion(+)
monitor.c +1 -0
index 34cee74..7857300 100644
@@ -5252,6 +5252,7 @@ static void monitor_event(void *opaque, int event)
         break;
 
     case CHR_EVENT_CLOSED:
+        readline_restart(mon->rs);
         mon_refcount--;
         monitor_fdsets_cleanup();
         break;
Signature · 1 line
1.7.10.4
#2 lore raw

Re: [Qemu-devel] [PATCH 1/2] monitor: Reset HMP mon->rs on CHR_EVENT_CLOSED

Quoted 5 lines · Stratos Psomadakis

Commit cdaa86a54 ("Add G_IO_HUP handler for socket chardev") exposed a bug in the way the HMP monitor handles its input. When a client closes the connection to the monitor, tcp_chr_read() will catch the HUP 'signal' and call tcp_chr_disconnect() to close the server-side connection too.

Your wording suggests SIGUP, but that's misleading. Suggest "tcp_chr_read() will detect the G_IO_HUP condition, and call".

Quoted 3 lines

Due to the fact that monitor reads 1 byte at a time (for
each tcp_chr_read()), the monitor readline state / buffers can be left in an inconsistent state (i.e. a half-finished command).

The state is not really inconsistent, there's just junk left in rs->cmd_buf[].

Quoted 3 lines

Thus, without
calling readline_restart() on mon->rs upon CHR_EVENT_CLOSED, future HMP commands will fail.

To make sure I understand you correctly: when you connect again, any leftover junk is prepended to your input, which messes up your first command. Correct?

Quoted 11 lines
---
 monitor.c |    1 +
 1 file changed, 1 insertion(+)
monitor.c +0 -0
index 34cee74..7857300 100644

monitor.c static void monitor_event(void *opaque, int event) · line 5258

Quoted 7 lines
monitor.c +1 -0
@@ -5252,6 +5252,7 @@ static void monitor_event(void *opaque, int event)
         break;
 
     case CHR_EVENT_CLOSED:
+        readline_restart(mon->rs);
         mon_refcount--;
         monitor_fdsets_cleanup();
         break;

Patch looks good to me.

#3 lore raw

Re: [Qemu-devel] [synnefo-devel] Re: [PATCH 1/2] monitor: Reset HMP mon->rs on CHR_EVENT_CLOSED

Quoted 9 lines · Markus Armbruster
Quoted 5 lines · Stratos Psomadakis

Commit cdaa86a54 ("Add G_IO_HUP handler for socket chardev") exposed a bug in the way the HMP monitor handles its input. When a client closes the connection to the monitor, tcp_chr_read() will catch the HUP 'signal' and call tcp_chr_disconnect() to close the server-side connection too.

Your wording suggests SIGUP, but that's misleading. Suggest "tcp_chr_read() will detect the G_IO_HUP condition, and call".

ack

Quoted 6 lines
Quoted 4 lines

Due to the fact that monitor reads 1 byte at a time (for
each tcp_chr_read()), the monitor readline state / buffers can be left in an inconsistent state (i.e. a half-finished command).

The state is not really inconsistent, there's just junk left in rs->cmd_buf[].

ack

Quoted 6 lines
Quoted 3 lines

Thus, without
calling readline_restart() on mon->rs upon CHR_EVENT_CLOSED, future HMP commands will fail.

To make sure I understand you correctly: when you connect again, any leftover junk is prepended to your input, which messes up your first command. Correct?

Yeap.

Quoted 13 lines
Quoted 13 lines
---
 monitor.c |    1 +
 1 file changed, 1 insertion(+)
monitor.c +0 -0
index 34cee74..7857300 100644
@@ -5252,6 +5252,7 @@ static void monitor_event(void *opaque, int event)
         break;
 
Quoted 6 lines

monitor.c static void monitor_event(void *opaque, int event) · line 5258

Quoted 5 lines
monitor.c +1 -0
     case CHR_EVENT_CLOSED:
+        readline_restart(mon->rs);
         mon_refcount--;
         monitor_fdsets_cleanup();
         break;

Patch looks good to me.

ok, I'll edit the commit msg and resend the patch.

Thanks,
Stratos

Quoted 1 line
Signature · 2 lines
Stratos Psomadakis
<[email protected]>
#4 lore raw

PATCH RESEND 1/2 monitor: Reset HMP mon->rs on CHR_EVENT_CLOSED

In reply to Markus Armbruster

Commit cdaa86a54 ("Add G_IO_HUP handler for socket chardev") exposed a bug in the way the HMP monitor handles its command buffer. When a client closes the connection to the monitor, tcp_chr_read() will detect the G_IO_HUP condition and call tcp_chr_disconnect() to close the server-side connection too. Due to the fact that monitor reads 1 byte at a time (for each tcp_chr_read()), the monitor readline state / buffers might contain junk (i.e. a half-finished command). Thus, without calling readline_restart() on mon->rs upon CHR_EVENT_CLOSED, future HMP commands will fail.

---
 monitor.c |    1 +
 1 file changed, 1 insertion(+)
monitor.c +1 -0
index 34cee74..7857300 100644
@@ -5252,6 +5252,7 @@ static void monitor_event(void *opaque, int event)
         break;
 
     case CHR_EVENT_CLOSED:
+        readline_restart(mon->rs);
         mon_refcount--;
         monitor_fdsets_cleanup();
         break;
Signature · 1 line
1.7.10.4
#5 lore raw

Re: [Qemu-devel] [PATCH resend 1/2] monitor: Reset HMP mon->rs on CHR_EVENT_CLOSED

Quoted 8 lines · On Fri, 12 Sep 2014 17:07:32 +0300 Stratos Psomadakis <[email protected]> wrote:

Commit cdaa86a54 ("Add G_IO_HUP handler for socket chardev") exposed a bug in the way the HMP monitor handles its command buffer. When a client closes the connection to the monitor, tcp_chr_read() will detect the G_IO_HUP condition and call tcp_chr_disconnect() to close the server-side connection too. Due to the fact that monitor reads 1 byte at a time (for each tcp_chr_read()), the monitor readline state / buffers might contain junk (i.e. a half-finished command). Thus, without calling readline_restart() on mon->rs upon CHR_EVENT_CLOSED, future HMP commands will fail.

What's your reproducer? Are you using the mux feature? We also reset it in CHR_EVENT_OPENED if the mux feature is not used, why isn't that good enough?

Quoted 19 lines
---
 monitor.c |    1 +
 1 file changed, 1 insertion(+)
monitor.c +1 -0
index 34cee74..7857300 100644
@@ -5252,6 +5252,7 @@ static void monitor_event(void *opaque, int event)
         break;
 
     case CHR_EVENT_CLOSED:
+        readline_restart(mon->rs);
         mon_refcount--;
         monitor_fdsets_cleanup();
         break;
#6 lore raw

Re: [Qemu-devel] [synnefo-devel] Re: [PATCH resend 1/2] monitor: Reset HMP mon->rs on CHR_EVENT_CLOSED

Quoted 11 lines · Luiz Capitulino
Quoted 8 lines · On Fri, 12 Sep 2014 17:07:32 +0300 Stratos Psomadakis <[email protected]> wrote:

Commit cdaa86a54 ("Add G_IO_HUP handler for socket chardev") exposed a bug in the way the HMP monitor handles its command buffer. When a client closes the connection to the monitor, tcp_chr_read() will detect the G_IO_HUP condition and call tcp_chr_disconnect() to close the server-side connection too. Due to the fact that monitor reads 1 byte at a time (for each tcp_chr_read()), the monitor readline state / buffers might contain junk (i.e. a half-finished command). Thus, without calling readline_restart() on mon->rs upon CHR_EVENT_CLOSED, future HMP commands will fail.

Quoted 1 line

What's your reproducer?

We have a script that opens a connection to the HMP socket and starts sending 'info version' commands to the monitor in a loop. If we kill the script (in the middle of the loop) and re-run it, we get "unknown command" errors from the HMP ("unknown command: 'infinfo'" for example).

Quoted 1 line

Are you using the mux feature?

Nope (on the cli we use '-monitor unix:<path>.mon,server,nowait' for the HMP).

Quoted 3 lines

We also reset it
in CHR_EVENT_OPENED if the mux feature is not used, why isn't that good enough?

I checked the code and on CHR_EVENT_OPENED the monitor calls readline_show_prompt (when not using mux). This resets the last_cmd_index/size readline variables, but the cmd_buf_index/size remains intact. I think that readline_restart() is necessary in order to cleanup the readline cmd buf (either in CHR_EVENT_OPENED or in CHR_EVENT_CLOSED).

Thanks,
Stratos

Quoted 19 lines
Quoted 19 lines
---
 monitor.c |    1 +
 1 file changed, 1 insertion(+)
monitor.c +1 -0
index 34cee74..7857300 100644
@@ -5252,6 +5252,7 @@ static void monitor_event(void *opaque, int event)
         break;
 
     case CHR_EVENT_CLOSED:
+        readline_restart(mon->rs);
         mon_refcount--;
         monitor_fdsets_cleanup();
         break;
Signature · 2 lines
Stratos Psomadakis
<[email protected]>
#7 lore raw

Re: [Qemu-devel] [synnefo-devel] Re: [PATCH resend 1/2] monitor: Reset HMP mon->rs on CHR_EVENT_CLOSED

Quoted 28 lines · On Fri, 12 Sep 2014 20:01:04 +0300 Stratos Psomadakis <[email protected]> wrote:
Quoted 11 lines · Luiz Capitulino
Quoted 8 lines · On Fri, 12 Sep 2014 17:07:32 +0300 Stratos Psomadakis <[email protected]> wrote:

Commit cdaa86a54 ("Add G_IO_HUP handler for socket chardev") exposed a bug in the way the HMP monitor handles its command buffer. When a client closes the connection to the monitor, tcp_chr_read() will detect the G_IO_HUP condition and call tcp_chr_disconnect() to close the server-side connection too. Due to the fact that monitor reads 1 byte at a time (for each tcp_chr_read()), the monitor readline state / buffers might contain junk (i.e. a half-finished command). Thus, without calling readline_restart() on mon->rs upon CHR_EVENT_CLOSED, future HMP commands will fail.

Quoted 1 line

What's your reproducer?

We have a script that opens a connection to the HMP socket and starts sending 'info version' commands to the monitor in a loop. If we kill the script (in the middle of the loop) and re-run it, we get "unknown command" errors from the HMP ("unknown command: 'infinfo'" for example).

Quoted 1 line

Are you using the mux feature?

Nope (on the cli we use '-monitor unix:<path>.mon,server,nowait' for the HMP).

Quoted 3 lines

We also reset it
in CHR_EVENT_OPENED if the mux feature is not used, why isn't that good enough?

Quoted 6 lines

I checked the code and on CHR_EVENT_OPENED the monitor calls readline_show_prompt (when not using mux). This resets the last_cmd_index/size readline variables, but the cmd_buf_index/size remains intact. I think that readline_restart() is necessary in order to cleanup the readline cmd buf (either in CHR_EVENT_OPENED or in CHR_EVENT_CLOSED).

I'm wondering if calling readline_restart() in the CHR_EVENT_CLOSED can break mux support. But I won't have time to check it today. Maybe moving the readline_restart() call to right before the readline_show_prompt() call in the OPENED event is the best thing to do?

Quoted 25 lines

Thanks,
Stratos

Quoted 19 lines
Quoted 19 lines
---
 monitor.c |    1 +
 1 file changed, 1 insertion(+)
monitor.c +1 -0
index 34cee74..7857300 100644
@@ -5252,6 +5252,7 @@ static void monitor_event(void *opaque, int event)
         break;
 
     case CHR_EVENT_CLOSED:
+        readline_restart(mon->rs);
         mon_refcount--;
         monitor_fdsets_cleanup();
         break;
#8 lore raw

Re: [Qemu-devel] [synnefo-devel] Re: [PATCH resend 1/2] monitor: Reset HMP mon->rs on CHR_EVENT_CLOSED

Quoted 34 lines · Luiz Capitulino
Quoted 25 lines · On Fri, 12 Sep 2014 20:01:04 +0300 Stratos Psomadakis <[email protected]> wrote:
Quoted 11 lines · Luiz Capitulino
Quoted 8 lines · On Fri, 12 Sep 2014 17:07:32 +0300 Stratos Psomadakis <[email protected]> wrote:

Commit cdaa86a54 ("Add G_IO_HUP handler for socket chardev") exposed a bug in the way the HMP monitor handles its command buffer. When a client closes the connection to the monitor, tcp_chr_read() will detect the G_IO_HUP condition and call tcp_chr_disconnect() to close the server-side connection too. Due to the fact that monitor reads 1 byte at a time (for each tcp_chr_read()), the monitor readline state / buffers might contain junk (i.e. a half-finished command). Thus, without calling readline_restart() on mon->rs upon CHR_EVENT_CLOSED, future HMP commands will fail.

Quoted 1 line

What's your reproducer?

We have a script that opens a connection to the HMP socket and starts sending 'info version' commands to the monitor in a loop. If we kill the script (in the middle of the loop) and re-run it, we get "unknown command" errors from the HMP ("unknown command: 'infinfo'" for example).

Quoted 1 line

Are you using the mux feature?

Nope (on the cli we use '-monitor unix:<path>.mon,server,nowait' for the HMP).

Quoted 3 lines

We also reset it
in CHR_EVENT_OPENED if the mux feature is not used, why isn't that good enough?

Quoted 6 lines

I checked the code and on CHR_EVENT_OPENED the monitor calls readline_show_prompt (when not using mux). This resets the last_cmd_index/size readline variables, but the cmd_buf_index/size remains intact. I think that readline_restart() is necessary in order to cleanup the readline cmd buf (either in CHR_EVENT_OPENED or in CHR_EVENT_CLOSED).

Quoted 4 lines

I'm wondering if calling readline_restart() in the CHR_EVENT_CLOSED can break mux support. But I won't have time to check it today. Maybe moving the readline_restart() call to right before the readline_show_prompt() call in the OPENED event is the best thing to do?

I did some quick tests with a mux chardev (I tried two mux'ed HMP monitors and a serial and an HMP). Calling readline_restart() in CHR_EVENT_CLOSED didn't seem to affect mux support (as far as I could tell). However, calling readline_restart() in CHR_EVENT_OPENED, just before readline_show_prompt(), resolves the issue too, and I think it makes more sense to be called at that point. If you agree, I can resend the modified patch.

Quoted 23 lines
Quoted 23 lines

Thanks,
Stratos

Quoted 18 lines
Quoted 18 lines
---
 monitor.c |    1 +
 1 file changed, 1 insertion(+)
monitor.c +1 -0
index 34cee74..7857300 100644
@@ -5252,6 +5252,7 @@ static void monitor_event(void *opaque, int event)
         break;
 
     case CHR_EVENT_CLOSED:
+        readline_restart(mon->rs);
         mon_refcount--;
         monitor_fdsets_cleanup();
         break;
Signature · 2 lines
Stratos Psomadakis
<[email protected]>
#9 lore raw

Re: [Qemu-devel] [synnefo-devel] Re: [PATCH resend 1/2] monitor: Reset HMP mon->rs on CHR_EVENT_CLOSED

Quoted 40 lines · On Sat, 13 Sep 2014 19:27:46 +0300 Stratos Psomadakis <[email protected]> wrote:
Quoted 34 lines · Luiz Capitulino
Quoted 25 lines · On Fri, 12 Sep 2014 20:01:04 +0300 Stratos Psomadakis <[email protected]> wrote:
Quoted 11 lines · Luiz Capitulino
Quoted 8 lines · On Fri, 12 Sep 2014 17:07:32 +0300 Stratos Psomadakis <[email protected]> wrote:

Commit cdaa86a54 ("Add G_IO_HUP handler for socket chardev") exposed a bug in the way the HMP monitor handles its command buffer. When a client closes the connection to the monitor, tcp_chr_read() will detect the G_IO_HUP condition and call tcp_chr_disconnect() to close the server-side connection too. Due to the fact that monitor reads 1 byte at a time (for each tcp_chr_read()), the monitor readline state / buffers might contain junk (i.e. a half-finished command). Thus, without calling readline_restart() on mon->rs upon CHR_EVENT_CLOSED, future HMP commands will fail.

Quoted 1 line

What's your reproducer?

We have a script that opens a connection to the HMP socket and starts sending 'info version' commands to the monitor in a loop. If we kill the script (in the middle of the loop) and re-run it, we get "unknown command" errors from the HMP ("unknown command: 'infinfo'" for example).

Quoted 1 line

Are you using the mux feature?

Nope (on the cli we use '-monitor unix:<path>.mon,server,nowait' for the HMP).

Quoted 3 lines

We also reset it
in CHR_EVENT_OPENED if the mux feature is not used, why isn't that good enough?

Quoted 6 lines

I checked the code and on CHR_EVENT_OPENED the monitor calls readline_show_prompt (when not using mux). This resets the last_cmd_index/size readline variables, but the cmd_buf_index/size remains intact. I think that readline_restart() is necessary in order to cleanup the readline cmd buf (either in CHR_EVENT_OPENED or in CHR_EVENT_CLOSED).

Quoted 4 lines

I'm wondering if calling readline_restart() in the CHR_EVENT_CLOSED can break mux support. But I won't have time to check it today. Maybe moving the readline_restart() call to right before the readline_show_prompt() call in the OPENED event is the best thing to do?

Quoted 7 lines

I did some quick tests with a mux chardev (I tried two mux'ed HMP monitors and a serial and an HMP). Calling readline_restart() in CHR_EVENT_CLOSED didn't seem to affect mux support (as far as I could tell). However, calling readline_restart() in CHR_EVENT_OPENED, just before readline_show_prompt(), resolves the issue too, and I think it makes more sense to be called at that point. If you agree, I can resend the modified patch.

Yes, I think that's the best. I'll just apply your respin.

Quoted 26 lines
Quoted 24 lines
Quoted 24 lines

Thanks,
Stratos

Quoted 18 lines
Quoted 18 lines
---
 monitor.c |    1 +
 1 file changed, 1 insertion(+)
monitor.c +1 -0
index 34cee74..7857300 100644
@@ -5252,6 +5252,7 @@ static void monitor_event(void *opaque, int event)
         break;
 
     case CHR_EVENT_CLOSED:
+        readline_restart(mon->rs);
         mon_refcount--;
         monitor_fdsets_cleanup();
         break;
#10 lore raw

PATCH RESEND v2 1/2 monitor: Reset HMP mon->rs in CHR_EVENT_OPEN

Commit cdaa86a54 ("Add G_IO_HUP handler for socket chardev") exposed a bug in the way the HMP monitor handles its command buffer. When a client closes the connection to the monitor, tcp_chr_read() will detect the G_IO_HUP condition and call tcp_chr_disconnect() to close the server-side connection too. Due to the fact that monitor reads 1 byte at a time (for each tcp_chr_read()), the monitor readline state / buffers might contain junk (i.e. a half-finished command). Thus, without calling readline_restart() on mon->rs in CHR_EVENT_OPEN, future HMP commands will fail.

---
 monitor.c |    1 +
 1 file changed, 1 insertion(+)
monitor.c +1 -0
index 34cee74..fb266bc 100644
@@ -5245,6 +5245,7 @@ static void monitor_event(void *opaque, int event)
         monitor_printf(mon, "QEMU %s monitor - type 'help' for more "
                        "information\n", QEMU_VERSION);
         if (!mon->mux_out) {
+            readline_restart(mon->rs);
             readline_show_prompt(mon->rs);
         }
         mon->reset_seen = 1;
Signature · 1 line
1.7.10.4
#11 lore raw

Re: [Qemu-devel] [PATCH resend v2 1/2] monitor: Reset HMP mon->rs in CHR_EVENT_OPEN

Quoted 11 lines · On Mon, 15 Sep 2014 15:34:57 +0300 Stratos Psomadakis <[email protected]> wrote:

Commit cdaa86a54 ("Add G_IO_HUP handler for socket chardev") exposed a bug in the way the HMP monitor handles its command buffer. When a client closes the connection to the monitor, tcp_chr_read() will detect the G_IO_HUP condition and call tcp_chr_disconnect() to close the server-side connection too. Due to the fact that monitor reads 1 byte at a time (for each tcp_chr_read()), the monitor readline state / buffers might contain junk (i.e. a half-finished command). Thus, without calling readline_restart() on mon->rs in CHR_EVENT_OPEN, future HMP commands will fail.

Applied to the qmp branch, thanks.

Quoted 16 lines
---
 monitor.c |    1 +
 1 file changed, 1 insertion(+)
monitor.c +1 -0
index 34cee74..fb266bc 100644
@@ -5245,6 +5245,7 @@ static void monitor_event(void *opaque, int event)
         monitor_printf(mon, "QEMU %s monitor - type 'help' for more "
                        "information\n", QEMU_VERSION);
         if (!mon->mux_out) {
+            readline_restart(mon->rs);
             readline_show_prompt(mon->rs);
         }
         mon->reset_seen = 1;
#12 lore raw

PATCH 2/2 iotests: Send the correct fd in socket_scm_helper

In reply to Stratos Psomadakis

Make sure to pass the correct fd via SCM_RIGHTS in socket_scm_helper.c (i.e. fd_to_send, not socket-fd).

---
 tests/qemu-iotests/socket_scm_helper.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
tests/qemu-iotests/socket_scm_helper.c +1 -1
index 0e2b285..8195983 100644
@@ -52,7 +52,7 @@ static int send_fd(int fd, int fd_to_send)
     cmsg->cmsg_len = CMSG_LEN(sizeof(int));
     cmsg->cmsg_level = SOL_SOCKET;
     cmsg->cmsg_type = SCM_RIGHTS;
-    memcpy(CMSG_DATA(cmsg), &fd, sizeof(int));
+    memcpy(CMSG_DATA(cmsg), &fd_to_send, sizeof(int));
 
     do {
         ret = sendmsg(fd, &msg, 0);
Signature · 1 line
1.7.10.4
#13 lore raw

Re: [Qemu-devel] [PATCH 2/2] iotests: Send the correct fd in socket_scm_helper

Quoted 14 lines · Stratos Psomadakis

Make sure to pass the correct fd via SCM_RIGHTS in socket_scm_helper.c (i.e. fd_to_send, not socket-fd).

---
 tests/qemu-iotests/socket_scm_helper.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
tests/qemu-iotests/socket_scm_helper.c +0 -0
index 0e2b285..8195983 100644

tests/qemu-iotests/socket_scm_helper.c static int send_fd(int fd, int fd_to_send) · line 58

Quoted 8 lines
tests/qemu-iotests/socket_scm_helper.c +1 -1
@@ -52,7 +52,7 @@ static int send_fd(int fd, int fd_to_send)
     cmsg->cmsg_len = CMSG_LEN(sizeof(int));
     cmsg->cmsg_level = SOL_SOCKET;
     cmsg->cmsg_type = SCM_RIGHTS;
-    memcpy(CMSG_DATA(cmsg), &fd, sizeof(int));
+    memcpy(CMSG_DATA(cmsg), &fd_to_send, sizeof(int));
 
     do {
         ret = sendmsg(fd, &msg, 0);

Ouch. Do you have an idea what's broken without this fix?

Applied
#14 lore raw

Re: [Qemu-devel] [PATCH 2/2] iotests: Send the correct fd in socket_scm_helper

Am 12.09.2014 um 09:04 hat Markus Armbruster geschrieben:

Quoted 7 lines
Quoted 5 lines · Stratos Psomadakis

Make sure to pass the correct fd via SCM_RIGHTS in socket_scm_helper.c (i.e. fd_to_send, not socket-fd).

Thanks, applied to the block branch.

(Also thanks to Markus for copying me, would have missed the patch otherwise.)

Quoted 14 lines
Quoted 14 lines
 tests/qemu-iotests/socket_scm_helper.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
tests/qemu-iotests/socket_scm_helper.c +1 -1
index 0e2b285..8195983 100644
@@ -52,7 +52,7 @@ static int send_fd(int fd, int fd_to_send)
     cmsg->cmsg_len = CMSG_LEN(sizeof(int));
     cmsg->cmsg_level = SOL_SOCKET;
     cmsg->cmsg_type = SCM_RIGHTS;
-    memcpy(CMSG_DATA(cmsg), &fd, sizeof(int));
+    memcpy(CMSG_DATA(cmsg), &fd_to_send, sizeof(int));
 
Quoted 4 lines

tests/qemu-iotests/socket_scm_helper.c static int send_fd(int fd, int fd_to_send) · line 58

Quoted 2 lines
tests/qemu-iotests/socket_scm_helper.c +0 -0
     do {
         ret = sendmsg(fd, &msg, 0);

Ouch. Do you have an idea what's broken without this fix?

As far as I can tell, nothing. Test case 045 will send a different file descriptor than it intended to, but the file descriptors aren't used other than checking whether qemu correctly reports their existence, so it doesn't matter.

I'm not adding qemu-stable therefore. Please correct me if I'm missing something.

Kevin

#15 lore raw

Re: [Qemu-devel] [PATCH 2/2] iotests: Send the correct fd in socket_scm_helper

Quoted 30 lines · Kevin Wolf

Am 12.09.2014 um 09:04 hat Markus Armbruster geschrieben:

Quoted 7 lines
Quoted 5 lines · Stratos Psomadakis

Make sure to pass the correct fd via SCM_RIGHTS in socket_scm_helper.c (i.e. fd_to_send, not socket-fd).

Thanks, applied to the block branch.

(Also thanks to Markus for copying me, would have missed the patch otherwise.)

Quoted 14 lines
Quoted 14 lines
 tests/qemu-iotests/socket_scm_helper.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
tests/qemu-iotests/socket_scm_helper.c +1 -1
index 0e2b285..8195983 100644
@@ -52,7 +52,7 @@ static int send_fd(int fd, int fd_to_send)
     cmsg->cmsg_len = CMSG_LEN(sizeof(int));
     cmsg->cmsg_level = SOL_SOCKET;
     cmsg->cmsg_type = SCM_RIGHTS;
-    memcpy(CMSG_DATA(cmsg), &fd, sizeof(int));
+    memcpy(CMSG_DATA(cmsg), &fd_to_send, sizeof(int));
 
Quoted 3 lines

tests/qemu-iotests/socket_scm_helper.c static int send_fd(int fd, int fd_to_send) · line 58

Quoted 2 lines
tests/qemu-iotests/socket_scm_helper.c +0 -0
     do {
         ret = sendmsg(fd, &msg, 0);

Ouch. Do you have an idea what's broken without this fix?

Quoted 7 lines

As far as I can tell, nothing. Test case 045 will send a different file descriptor than it intended to, but the file descriptors aren't used other than checking whether qemu correctly reports their existence, so it doesn't matter.

I'm not adding qemu-stable therefore. Please correct me if I'm missing something.

Right. I mentioned qemu-stable mainly for the first patch.

Thanks,
Stratos

Quoted 2 lines

Kevin

Signature · 2 lines
Stratos Psomadakis
<[email protected]>
#16 lore raw
In reply to Stratos Psomadakis

You neglected to cc maintainers.

Quoted 4 lines · Stratos Psomadakis

Hi,

the first patch fixes an issue with HMP monitors, which was exposed with v2.1.0 (commits cdaa86a and 812c10).

Copying Luiz.

Quoted 2 lines

The second one fixes a typo in a helper C program used in qemu-iotests.

Copying Kevin and Stefan.

Quoted 1 line

We think that they should be cherry-picked for the next stable release.

Copying [email protected].