linux-kernel - Re: [PATCH v2 2/6] perf annotate: Simplify disasm

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Tue, 28 Jan 2020 19:27:29 +0530
From:   Ravi Bangoria <ravi.bangoria@...ux.ibm.com>
To:     Jiri Olsa <jolsa@...hat.com>
Cc:     acme@...nel.org, namhyung@...nel.org, irogers@...gle.com,
        songliubraving@...com, yao.jin@...ux.intel.com,
        linux-kernel@...r.kernel.org,
        Ravi Bangoria <ravi.bangoria@...ux.ibm.com>
Subject: Re: [PATCH v2 2/6] perf annotate: Simplify disasm_line allocation and
 freeing code

Hi Jiri,

On 1/27/20 5:29 PM, Jiri Olsa wrote:
> On Fri, Jan 24, 2020 at 01:34:28PM +0530, Ravi Bangoria wrote:
> 
> SNIP
> 
>>   
>>   /*
>>    * Allocating the disasm annotation line data with
>>    * following structure:
>>    *
>> - *    ------------------------------------------------------------
>> - *    privsize space | struct disasm_line | struct annotation_line
>> - *    ------------------------------------------------------------
>> + *    -------------------------------------------
>> + *    struct disasm_line | struct annotation_line
>> + *    -------------------------------------------
>>    *
>>    * We have 'struct annotation_line' member as last member
>>    * of 'struct disasm_line' to have an easy access.
>> - *
>>    */
>>   static struct disasm_line *disasm_line__new(struct annotate_args *args)
>>   {
>>   	struct disasm_line *dl = NULL;
>> -	struct annotation_line *al;
>> -	size_t privsize = args->privsize + offsetof(struct disasm_line, al);
>> +	int nr = 1;
>>   
>> -	al = annotation_line__new(args, privsize);
> 
> ok, I finally recalled why we did it like this.. for the python
> annotation support, which never made it in ;-) however the allocation
> in 'specialized' line and later call to annotation_line__init might
> actualy be a better way

Sorry, I didn't get your point about 'specialized' line. You mean the way
I'm doing is better wrt existing code?

> 
>> -	if (al != NULL) {
>> -		dl = disasm_line(al);
>> +	if (perf_evsel__is_group_event(args->evsel))
>> +		nr = args->evsel->core.nr_members;
>>   
>> -		if (dl->al.line == NULL)
>> -			goto out_delete;
>> +	dl = zalloc(disasm_line_size(nr));
>> +	if (!dl)
>> +		return NULL;
>>   
>> -		if (args->offset != -1) {
>> -			if (disasm_line__parse(dl->al.line, &dl->ins.name, &dl->ops.raw) < 0)
>> -				goto out_free_line;
>> +	annotation_line__init(&dl->al, args, nr);
>> +	if (dl->al.line == NULL)
>> +		goto out_delete;
>>   
>> -			disasm_line__init_ins(dl, args->arch, &args->ms);
>> -		}
>> +	if (args->offset != -1) {
>> +		if (disasm_line__parse(dl->al.line, &dl->ins.name, &dl->ops.raw) < 0)
>> +			goto out_free_line;
>> +
>> +		disasm_line__init_ins(dl, args->arch, &args->ms);
>>   	}
>>   
>>   	return dl;
>> @@ -1248,7 +1219,9 @@ void disasm_line__free(struct disasm_line *dl)
>>   	else
>>   		ins__delete(&dl->ops);
>>   	zfree(&dl->ins.name);
>> -	annotation_line__delete(&dl->al);
>> +	free_srcline(dl->al.path);
>> +	zfree(&dl->al.line);
> 
> no need to zfree if you're freeing the memory on the next line

dl->al.line is allocated using strdup() in annotation_line__init(). So I
think we still need that zfree(&dl->al.line).

> also could you please put it to annotation_line__exit, since
> you already added the __init function

Sure will do it.

Thanks for review :)
Ravi