* [PATCH] qemu-nbd: Fix a memleak in nbd_client_thread()
@ 2020-12-01 6:13 Alex Chen
2020-12-01 20:15 ` Eric Blake
0 siblings, 1 reply; 4+ messages in thread
From: Alex Chen @ 2020-12-01 6:13 UTC (permalink / raw)
To: eblake
Cc: alex.chen, qemu-trivial, qemu-devel, qemu-block, zhang.zhanghailiang
When the qio_channel_socket_connect_sync() fails
we should goto 'out_socket' label to free the 'sioc' instead of
goto 'out' label.
In addition, now the 'out' label is useless, delete it.
Reported-by: Euler Robot <euler.robot@huawei.com>
Signed-off-by: Alex Chen <alex.chen@huawei.com>
---
qemu-nbd.c | 3 +--
1 file changed, 1 insertion(+), 2 deletions(-)
diff --git a/qemu-nbd.c b/qemu-nbd.c
index 47587a709e..643b0777c0 100644
--- a/qemu-nbd.c
+++ b/qemu-nbd.c
@@ -275,7 +275,7 @@ static void *nbd_client_thread(void *arg)
saddr,
&local_error) < 0) {
error_report_err(local_error);
- goto out;
+ goto out_socket;
}
ret = nbd_receive_negotiate(NULL, QIO_CHANNEL(sioc),
@@ -325,7 +325,6 @@ out_fd:
close(fd);
out_socket:
object_unref(OBJECT(sioc));
-out:
g_free(info.name);
kill(getpid(), SIGTERM);
return (void *) EXIT_FAILURE;
--
2.19.1
^ permalink raw reply related [flat|nested] 4+ messages in thread
* Re: [PATCH] qemu-nbd: Fix a memleak in nbd_client_thread()
2020-12-01 6:13 [PATCH] qemu-nbd: Fix a memleak in nbd_client_thread() Alex Chen
@ 2020-12-01 20:15 ` Eric Blake
2020-12-02 2:02 ` Alex Chen
0 siblings, 1 reply; 4+ messages in thread
From: Eric Blake @ 2020-12-01 20:15 UTC (permalink / raw)
To: Alex Chen; +Cc: qemu-trivial, qemu-devel, qemu-block, zhang.zhanghailiang
On 12/1/20 12:13 AM, Alex Chen wrote:
> When the qio_channel_socket_connect_sync() fails
> we should goto 'out_socket' label to free the 'sioc' instead of
> goto 'out' label.
> In addition, now the 'out' label is useless, delete it.
>
> Reported-by: Euler Robot <euler.robot@huawei.com>
> Signed-off-by: Alex Chen <alex.chen@huawei.com>
> ---
> qemu-nbd.c | 3 +--
> 1 file changed, 1 insertion(+), 2 deletions(-)
>
> diff --git a/qemu-nbd.c b/qemu-nbd.c
> index 47587a709e..643b0777c0 100644
> --- a/qemu-nbd.c
> +++ b/qemu-nbd.c
> @@ -275,7 +275,7 @@ static void *nbd_client_thread(void *arg)
> saddr,
> &local_error) < 0) {
> error_report_err(local_error);
> - goto out;
> + goto out_socket;
> }
>
> ret = nbd_receive_negotiate(NULL, QIO_CHANNEL(sioc),
> @@ -325,7 +325,6 @@ out_fd:
> close(fd);
> out_socket:
> object_unref(OBJECT(sioc));
> -out:
> g_free(info.name);
> kill(getpid(), SIGTERM);
> return (void *) EXIT_FAILURE;
>
While the patch looks correct, we have a lot of duplication. Simpler
might be a solution with only one exit label altogether:
diff --git i/qemu-nbd.c w/qemu-nbd.c
index a7075c5419d7..d7bdcd0011ba 100644
--- i/qemu-nbd.c
+++ w/qemu-nbd.c
@@ -265,8 +265,8 @@ static void *nbd_client_thread(void *arg)
char *device = arg;
NBDExportInfo info = { .request_sizes = false, .name = g_strdup("") };
QIOChannelSocket *sioc;
- int fd;
- int ret;
+ int fd = -1;
+ int ret = EXIT_FAILURE;
pthread_t show_parts_thread;
Error *local_error = NULL;
@@ -278,26 +278,24 @@ static void *nbd_client_thread(void *arg)
goto out;
}
- ret = nbd_receive_negotiate(NULL, QIO_CHANNEL(sioc),
- NULL, NULL, NULL, &info, &local_error);
- if (ret < 0) {
+ if (nbd_receive_negotiate(NULL, QIO_CHANNEL(sioc),
+ NULL, NULL, NULL, &info, &local_error) < 0) {
if (local_error) {
error_report_err(local_error);
}
- goto out_socket;
+ goto out;
}
fd = open(device, O_RDWR);
if (fd < 0) {
/* Linux-only, we can use %m in printf. */
error_report("Failed to open %s: %m", device);
- goto out_socket;
+ goto out;
}
- ret = nbd_init(fd, sioc, &info, &local_error);
- if (ret < 0) {
+ if (nbd_init(fd, sioc, &info, &local_error) < 0) {
error_report_err(local_error);
- goto out_fd;
+ goto out;
}
/* update partition table */
@@ -311,24 +309,18 @@ static void *nbd_client_thread(void *arg)
dup2(STDOUT_FILENO, STDERR_FILENO);
}
- ret = nbd_client(fd);
- if (ret) {
- goto out_fd;
+ if (nbd_client(fd) == 0) {
+ ret = EXIT_SUCCESS;
}
- close(fd);
- object_unref(OBJECT(sioc));
- g_free(info.name);
- kill(getpid(), SIGTERM);
- return (void *) EXIT_SUCCESS;
-out_fd:
- close(fd);
-out_socket:
+ out:
+ if (fd >= 0) {
+ close(fd);
+ }
object_unref(OBJECT(sioc));
-out:
g_free(info.name);
kill(getpid(), SIGTERM);
- return (void *) EXIT_FAILURE;
+ return (void *) (intptr_t) ret;
}
#endif /* HAVE_NBD_DEVICE */
--
Eric Blake, Principal Software Engineer
Red Hat, Inc. +1-919-301-3226
Virtualization: qemu.org | libvirt.org
^ permalink raw reply related [flat|nested] 4+ messages in thread
* Re: [PATCH] qemu-nbd: Fix a memleak in nbd_client_thread()
2020-12-01 20:15 ` Eric Blake
@ 2020-12-02 2:02 ` Alex Chen
2020-12-03 16:26 ` Eric Blake
0 siblings, 1 reply; 4+ messages in thread
From: Alex Chen @ 2020-12-02 2:02 UTC (permalink / raw)
To: Eric Blake; +Cc: qemu-trivial, qemu-devel, qemu-block, zhang.zhanghailiang
On 2020/12/2 4:15, Eric Blake wrote:
> On 12/1/20 12:13 AM, Alex Chen wrote:
>> When the qio_channel_socket_connect_sync() fails
>> we should goto 'out_socket' label to free the 'sioc' instead of
>> goto 'out' label.
>> In addition, now the 'out' label is useless, delete it.
>>
>> Reported-by: Euler Robot <euler.robot@huawei.com>
>> Signed-off-by: Alex Chen <alex.chen@huawei.com>
>> ---
>> qemu-nbd.c | 3 +--
>> 1 file changed, 1 insertion(+), 2 deletions(-)
>>
>> diff --git a/qemu-nbd.c b/qemu-nbd.c
>> index 47587a709e..643b0777c0 100644
>> --- a/qemu-nbd.c
>> +++ b/qemu-nbd.c
>> @@ -275,7 +275,7 @@ static void *nbd_client_thread(void *arg)
>> saddr,
>> &local_error) < 0) {
>> error_report_err(local_error);
>> - goto out;
>> + goto out_socket;
>> }
>>
>> ret = nbd_receive_negotiate(NULL, QIO_CHANNEL(sioc),
>> @@ -325,7 +325,6 @@ out_fd:
>> close(fd);
>> out_socket:
>> object_unref(OBJECT(sioc));
>> -out:
>> g_free(info.name);
>> kill(getpid(), SIGTERM);
>> return (void *) EXIT_FAILURE;
>>
>
> While the patch looks correct, we have a lot of duplication. Simpler
> might be a solution with only one exit label altogether:
>
Thanks for your review, I will modify the patch and send patch v2 according to your suggestion.
BTW, do I need to split this patch into two patches, one to solve the memleak and the other to optimizes the redundant code?
Thanks,
Alex
> diff --git i/qemu-nbd.c w/qemu-nbd.c
> index a7075c5419d7..d7bdcd0011ba 100644
> --- i/qemu-nbd.c
> +++ w/qemu-nbd.c
> @@ -265,8 +265,8 @@ static void *nbd_client_thread(void *arg)
> char *device = arg;
> NBDExportInfo info = { .request_sizes = false, .name = g_strdup("") };
> QIOChannelSocket *sioc;
> - int fd;
> - int ret;
> + int fd = -1;
> + int ret = EXIT_FAILURE;
> pthread_t show_parts_thread;
> Error *local_error = NULL;
>
> @@ -278,26 +278,24 @@ static void *nbd_client_thread(void *arg)
> goto out;
> }
>
> - ret = nbd_receive_negotiate(NULL, QIO_CHANNEL(sioc),
> - NULL, NULL, NULL, &info, &local_error);
> - if (ret < 0) {
> + if (nbd_receive_negotiate(NULL, QIO_CHANNEL(sioc),
> + NULL, NULL, NULL, &info, &local_error) < 0) {
> if (local_error) {
> error_report_err(local_error);
> }
> - goto out_socket;
> + goto out;
> }
>
> fd = open(device, O_RDWR);
> if (fd < 0) {
> /* Linux-only, we can use %m in printf. */
> error_report("Failed to open %s: %m", device);
> - goto out_socket;
> + goto out;
> }
>
> - ret = nbd_init(fd, sioc, &info, &local_error);
> - if (ret < 0) {
> + if (nbd_init(fd, sioc, &info, &local_error) < 0) {
> error_report_err(local_error);
> - goto out_fd;
> + goto out;
> }
>
> /* update partition table */
> @@ -311,24 +309,18 @@ static void *nbd_client_thread(void *arg)
> dup2(STDOUT_FILENO, STDERR_FILENO);
> }
>
> - ret = nbd_client(fd);
> - if (ret) {
> - goto out_fd;
> + if (nbd_client(fd) == 0) {
> + ret = EXIT_SUCCESS;
> }
> - close(fd);
> - object_unref(OBJECT(sioc));
> - g_free(info.name);
> - kill(getpid(), SIGTERM);
> - return (void *) EXIT_SUCCESS;
>
> -out_fd:
> - close(fd);
> -out_socket:
> + out:
> + if (fd >= 0) {
> + close(fd);
> + }
> object_unref(OBJECT(sioc));
> -out:
> g_free(info.name);
> kill(getpid(), SIGTERM);
> - return (void *) EXIT_FAILURE;
> + return (void *) (intptr_t) ret;
> }
> #endif /* HAVE_NBD_DEVICE */
>
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH] qemu-nbd: Fix a memleak in nbd_client_thread()
2020-12-02 2:02 ` Alex Chen
@ 2020-12-03 16:26 ` Eric Blake
0 siblings, 0 replies; 4+ messages in thread
From: Eric Blake @ 2020-12-03 16:26 UTC (permalink / raw)
To: Alex Chen; +Cc: qemu-trivial, qemu-devel, qemu-block, zhang.zhanghailiang
On 12/1/20 8:02 PM, Alex Chen wrote:
> On 2020/12/2 4:15, Eric Blake wrote:
>> While the patch looks correct, we have a lot of duplication. Simpler
>> might be a solution with only one exit label altogether:
>>
>
> Thanks for your review, I will modify the patch and send patch v2 according to your suggestion.
> BTW, do I need to split this patch into two patches, one to solve the memleak and the other to optimizes the redundant code?
This one seems small enough that a single patch is probably fine (that
is, document that the patch is fixing a memory loop BY consolidating
redundant code).
--
Eric Blake, Principal Software Engineer
Red Hat, Inc. +1-919-301-3226
Virtualization: qemu.org | libvirt.org
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2020-12-03 16:30 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-12-01 6:13 [PATCH] qemu-nbd: Fix a memleak in nbd_client_thread() Alex Chen
2020-12-01 20:15 ` Eric Blake
2020-12-02 2:02 ` Alex Chen
2020-12-03 16:26 ` Eric Blake
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).