[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [PATCH] pc: fix leak in pc_system_flash_cleanup_unused
From: |
Philippe Mathieu-Daudé |
Subject: |
Re: [PATCH] pc: fix leak in pc_system_flash_cleanup_unused |
Date: |
Wed, 1 Jul 2020 09:34:08 +0200 |
User-agent: |
Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101 Thunderbird/68.5.0 |
On 7/1/20 3:58 AM, Alexander Bulekov wrote:
> fix a leak detected when building with --enable-sanitizers:
> ./i386-softmmu/qemu-system-i386
> Upon exit:
> ==13576==ERROR: LeakSanitizer: detected memory leaks
>
> Direct leak of 1216 byte(s) in 1 object(s) allocated from:
> #0 0x7f9d2ed5c628 in malloc (/usr/lib/x86_64-linux-gnu/libasan.so.5)
> #1 0x7f9d2e963500 in g_malloc (/usr/lib/x86_64-linux-gnu/libglib-2.0.so.)
> #2 0x55fa646d25cc in object_new_with_type /tmp/qemu/qom/object.c:686
> #3 0x55fa63dbaa88 in qdev_new /tmp/qemu/hw/core/qdev.c:140
> #4 0x55fa638a533f in pc_pflash_create /tmp/qemu/hw/i386/pc_sysfw.c:88
> #5 0x55fa638a54c4 in pc_system_flash_create
> /tmp/qemu/hw/i386/pc_sysfw.c:106
> #6 0x55fa646caa1d in object_init_with_type /tmp/qemu/qom/object.c:369
> #7 0x55fa646d20b5 in object_initialize_with_type
> /tmp/qemu/qom/object.c:511
> #8 0x55fa646d2606 in object_new_with_type /tmp/qemu/qom/object.c:687
> #9 0x55fa639431e9 in qemu_init /tmp/qemu/softmmu/vl.c:3878
> #10 0x55fa6335c1b8 in main /tmp/qemu/softmmu/main.c:48
> #11 0x7f9d2cf06e0a in __libc_start_main ../csu/libc-start.c:308
> #12 0x55fa6335f8e9 in _start
> (/tmp/qemu/build/i386-softmmu/qemu-system-i386)
>
> Signed-off-by: Alexander Bulekov <alxndr@bu.edu>
> ---
>
> I am not very familiar with the QOM, so maybe this isn't the right way
> of going about this. With the call to object_property_add_child in
> pc_sysfw.c:pc_pflash_create, object_ref is called on the pflash device.
> In the pc_system_flash_cleanup_unused function, there are calls to
> object_propery_del and object_unparent, but it seems neither of these
> calls object_unref. So do we need to manually decrement the refcount?
Hint in such case, look at who wrote the commit, and Cc the author
to get an explanation:
$ git blame hw/i386/pc_sysfw.c
ebc29e1beab hw/i386/pc_sysfw.c (Markus Armbruster 2019-03-11
18:39:26 +0100 119) for (i = 0; i < ARRAY_SIZE(pcms->flash); i++) {
ebc29e1beab hw/i386/pc_sysfw.c (Markus Armbruster 2019-03-11
18:39:26 +0100 120) dev_obj = OBJECT(pcms->flash[i]);
ebc29e1beab hw/i386/pc_sysfw.c (Markus Armbruster 2019-03-11
18:39:26 +0100 121) if (!object_property_get_bool(dev_obj,
"realized", &error_abort)) {
ebc29e1beab hw/i386/pc_sysfw.c (Markus Armbruster 2019-03-11
18:39:26 +0100 122) prop_name = g_strdup_printf("pflash%d", i);
df4fe0b291b hw/i386/pc_sysfw.c (Markus Armbruster 2020-05-05
17:29:26 +0200 123) object_property_del(OBJECT(pcms),
prop_name);
ebc29e1beab hw/i386/pc_sysfw.c (Markus Armbruster 2019-03-11
18:39:26 +0100 124) g_free(prop_name);
ebc29e1beab hw/i386/pc_sysfw.c (Markus Armbruster 2019-03-11
18:39:26 +0100 125) object_unparent(dev_obj);
ebc29e1beab hw/i386/pc_sysfw.c (Markus Armbruster 2019-03-11
18:39:26 +0100 126) pcms->flash[i] = NULL;
ebc29e1beab hw/i386/pc_sysfw.c (Markus Armbruster 2019-03-11
18:39:26 +0100 127) }
ebc29e1beab hw/i386/pc_sysfw.c (Markus Armbruster 2019-03-11
18:39:26 +0100 128) }
>
> hw/i386/pc_sysfw.c | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/hw/i386/pc_sysfw.c b/hw/i386/pc_sysfw.c
> index ec2a3b3e7e..f69a93671a 100644
> --- a/hw/i386/pc_sysfw.c
> +++ b/hw/i386/pc_sysfw.c
> @@ -123,6 +123,7 @@ static void pc_system_flash_cleanup_unused(PCMachineState
> *pcms)
> object_property_del(OBJECT(pcms), prop_name);
> g_free(prop_name);
> object_unparent(dev_obj);
> + object_unref(dev_obj);
> pcms->flash[i] = NULL;
> }
> }
>