Re: [Qemu-devel] [PULL 9/9] colo-proxy: fix memory leak

[Jason Wang <jasowang@redhat.com>]

On 2016年10月31日 14:37, Zhang Chen wrote:
>
>
> On 10/30/2016 08:13 PM, Greg Kurz wrote:
>> On Wed, 26 Oct 2016 10:24:11 +0800
>> Jason Wang <jasowang@redhat.com> wrote:
>>
>>> From: Zhang Chen <zhangchen.fnst@cn.fujitsu.com>
>>>
>>> Fix memory leak in colo-compare.c and filter-rewriter.c
>>> Report by Coverity and add some comments.
>>>
>>> Signed-off-by: Zhang Chen <zhangchen.fnst@cn.fujitsu.com>
>>> Reviewed-by: zhanghailiang <zhang.zhanghailiang@huawei.com>
>>> Signed-off-by: Jason Wang <jasowang@redhat.com>
>>> ---
>> Hi,
>>
>> This commit breaks --enable-trace-backends=ust builds in travis (#17):
>>
>> In file included from ./trace/generated-tracers.h:5966:0,
>>                   from /home/travis/build/gkurz/qemu/include/trace.h:4,
>>                   from qapi/qapi-visit-core.c:22:
>> ./trace/generated-ust-provider.h:18625:3: error: unknown type name 
>> ‘_TP_EXPROTOint’
>> In file included from /home/travis/build/gkurz/qemu/include/trace.h:4:0,
>>                   from qapi/qapi-visit-core.c:22:
>> ./trace/generated-tracers.h: In function ‘trace_colo_compare_pkt_info’:
>> ./trace/generated-tracers.h:19249:432: error: expected string literal 
>> before ‘_SDT_ASM_OPERANDS_ssize’
>> ./trace/generated-tracers.h:19249:432: error: implicit declaration of 
>> function ‘__tracepoint_cb_qemu___colo_compare_pkt_info’ 
>> [-Werror=implicit-function-declaration]
>> ./trace/generated-tracers.h:19249:432: error: nested extern 
>> declaration of ‘__tracepoint_cb_qemu___colo_compare_pkt_info’ 
>> [-Werror=nested-externs]
>> cc1: all warnings being treated as errors
>> make: *** [qapi/qapi-visit-core.o] Error 1
>> make: *** Waiting for unfinished jobs....
>>
>> https://travis-ci.org/gkurz/qemu/jobs/171641119
>
> I got your point and test it, same problem in here.
> I found that when trace args >10 it not work(this trace have 11 args),
> else it works well like that:
>
> colo_compare_pkt_info(const char *src, const char *dst, uint32_t pseq, 
> uint32_t pack, uint32_t sseq, uint32_t sack, int res, uint32_t pflag, 
> uint32_t sflag, const char *src1) "src/dst: %s/%s p: seq/ack=%u/%u   
> s: seq/ack=%u/%u res=%d flags=%x/%x %s\n"
>
> So I think this bug could be related to trace system, but I review 
> trace codes nothing be found about this.
> CC: stefan
>
> Thanks 

We met similar issue in the past which looks like a limitation of the 
backend.

Btw, Alex Bennee has posted a workaround for this, could you please 
review or ack on that patch?

Thanks