[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Qemu-devel] [PATCH 05/18] migration: avoid using error_is_set
From: |
Luiz Capitulino |
Subject: |
Re: [Qemu-devel] [PATCH 05/18] migration: avoid using error_is_set |
Date: |
Thu, 4 Oct 2012 15:06:30 -0300 |
On Wed, 3 Oct 2012 16:36:52 +0200
Paolo Bonzini <address@hidden> wrote:
> Signed-off-by: Paolo Bonzini <address@hidden>
The patch's description is a bit misleading. The real problem this fixes
is that the migration code is using errp to detect "internal" errors,
this means that it relies on errp being non-NULL.
There's no problem using error_set(). Actually, I do prefer using it, because
at least in theory we shouldn't assume the Error object to have this
particular semantic.
One more comment below.
> ---
> migration-tcp.c | 8 +++++---
> migration.c | 13 +++++++------
> 2 file modificati, 12 inserzioni(+), 9 rimozioni(-)
>
> diff --git a/migration-tcp.c b/migration-tcp.c
> index a15c2b8..78337a3 100644
> --- a/migration-tcp.c
> +++ b/migration-tcp.c
> @@ -71,14 +71,16 @@ static void tcp_wait_for_connect(int fd, void *opaque)
> int tcp_start_outgoing_migration(MigrationState *s, const char *host_port,
> Error **errp)
> {
> + Error *local_err = NULL;
> +
> s->get_error = socket_errno;
> s->write = socket_write;
> s->close = tcp_close;
>
> - s->fd = inet_nonblocking_connect(host_port, tcp_wait_for_connect, s,
> - errp);
> - if (error_is_set(errp)) {
> + s->fd = inet_nonblocking_connect(host_port, tcp_wait_for_connect, s,
> &local_err);
> + if (local_err != NULL) {
> migrate_fd_error(s);
> + error_propagate(errp, local_err);
> return -1;
> }
>
> diff --git a/migration.c b/migration.c
> index 22a05c4..8a04174 100644
> --- a/migration.c
> +++ b/migration.c
> @@ -481,6 +481,7 @@ void qmp_migrate(const char *uri, bool has_blk, bool blk,
> bool has_inc, bool inc, bool has_detach, bool detach,
> Error **errp)
> {
> + Error *local_err = NULL;
> MigrationState *s = migrate_get_current();
> MigrationParams params;
> const char *p;
> @@ -506,7 +507,7 @@ void qmp_migrate(const char *uri, bool has_blk, bool blk,
> s = migrate_init(¶ms);
>
> if (strstart(uri, "tcp:", &p)) {
> - ret = tcp_start_outgoing_migration(s, p, errp);
> + ret = tcp_start_outgoing_migration(s, p, &local_err);
> #if !defined(WIN32)
> } else if (strstart(uri, "exec:", &p)) {
> ret = exec_start_outgoing_migration(s, p);
> @@ -520,11 +521,11 @@ void qmp_migrate(const char *uri, bool has_blk, bool
> blk,
> return;
> }
>
> - if (ret < 0) {
> - if (!error_is_set(errp)) {
> - DPRINTF("migration failed: %s\n", strerror(-ret));
> - /* FIXME: we should return meaningful errors */
> - error_set(errp, QERR_UNDEFINED_ERROR);
> + if (ret < 0 || local_err) {
> + if (!local_err) {
> + error_set_errno(errp, -ret, QERR_UNDEFINED_ERROR);
Two problems here. First, ret usually is not -errno. If we really want to
use it here (I think this is great improvement) than we have to fix the
functions called by qmp_migrate() first.
The other problem is just this will produce a weird error message, it's
better to do s/QERR_UNDEFINED_ERROR/"migration has failed".
> + } else {
> + error_propagate(errp, local_err);
> }
> return;
> }
- [Qemu-devel] [PATCH 02/18] qemu-sockets: add Error ** to all functions, (continued)
- [Qemu-devel] [PATCH 03/18] qemu-sockets: unix_listen and unix_connect are portable, Paolo Bonzini, 2012/10/03
- [Qemu-devel] [PATCH 05/18] migration: avoid using error_is_set, Paolo Bonzini, 2012/10/03
- Re: [Qemu-devel] [PATCH 05/18] migration: avoid using error_is_set,
Luiz Capitulino <=
- [Qemu-devel] [PATCH 04/18] qemu-sockets: add nonblocking connect for Unix sockets, Paolo Bonzini, 2012/10/03
- [Qemu-devel] [PATCH 06/18] migration: centralize call to migrate_fd_error(), Paolo Bonzini, 2012/10/03
- [Qemu-devel] [PATCH 07/18] migration: use qemu-sockets to establish Unix sockets, Paolo Bonzini, 2012/10/03
- [Qemu-devel] [PATCH 09/18] migration (incoming): add error propagation for fd and exec protocols, Paolo Bonzini, 2012/10/03
- [Qemu-devel] [PATCH 08/18] migration (outgoing): add error propagation for fd and exec protocols, Paolo Bonzini, 2012/10/03