Skip to contentSkip to conversation

Clean up error handling in net.c

PATCH In review Markus Armbruster 3 participants 4 messages Sep 28, 2009 – Sep 30, 2009

lore
Share & tools
lore
#0 lore raw

cover Clean up error handling in net.c

Caveat: I'm not familiar with slirp, and I'm not sure 1/3 is the right thing to do.

Markus Armbruster (3):
Make net_client_init() consume slirp_configs even on error Don't exit() in config_error()
Drop config_error(), use qemu_error() instead

 net.c |  192 ++++++++++++++++++++++++++++++++---------------------------------
 net.h |    4 +-
 vl.c  |    9 ++-
 3 files changed, 102 insertions(+), 103 deletions(-)
#1 lore raw
Quoted 2 lines · On Mon, 28 Sep 2009 21:11:32 +0200 Markus Armbruster <[email protected]> wrote:

Caveat: I'm not familiar with slirp, and I'm not sure 1/3 is the right thing to do.

FWIW, I'm working on a infrastructure for error handling in the Monitor.

The goal is not only to fix abort/don't abort cases, but also to pass
error information up in the call stack, so that the protocol can emit it.

So, in case you are planning to fix more of those abort cases, I would
really appreciate if you do this as part of a QObject conversion, otherwise I would have to re-visit the handler to do the conversion myself.

I should post patches this week, although I'm not sure if I've picked
up the best API design for this...

#2 lore raw
Quoted 8 lines · Luiz Capitulino
Quoted 2 lines · On Mon, 28 Sep 2009 21:11:32 +0200 Markus Armbruster <[email protected]> wrote:

Caveat: I'm not familiar with slirp, and I'm not sure 1/3 is the right thing to do.

FWIW, I'm working on a infrastructure for error handling in the Monitor.

Quoted 7 lines

The goal is not only to fix abort/don't abort cases, but also to pass
error information up in the call stack, so that the protocol can emit it.

So, in case you are planning to fix more of those abort cases, I would
really appreciate if you do this as part of a QObject conversion, otherwise I would have to re-visit the handler to do the conversion myself.

Understood.

This time around, I didn't get close to the handler. I just fixed low-level code called from both startup and monitor.

Quoted 2 lines

I should post patches this week, although I'm not sure if I've picked
up the best API design for this...

Yes, show us the code :)

#3 lore raw
In reply to Markus Armbruster
Quoted 7 lines · Markus Armbruster

Caveat: I'm not familiar with slirp, and I'm not sure 1/3 is the right thing to do.

Markus Armbruster (3):
Make net_client_init() consume slirp_configs even on error Don't exit() in config_error()
Drop config_error(), use qemu_error() instead

The series looks good to me

Since it conflicts significantly with my conversion of -net to QemuOpts, I've pulled them into my tree (modulo the spurious USB change) and re-based:

http://git.et.redhat.com/?p=qemu-net.git

Hope you agree that makes sense

Thanks,
Mark.