Re: [PATCH] perf build: Build error in libbpf missing initialization

[Daniel Borkmann]

On 07/27/2018 07:59 PM, Jakub Kicinski wrote:
> On Fri, 27 Jul 2018 10:21:26 +0200, Thomas Richter wrote:
>> In linux-next tree compiling the perf tool with additional make flags
>> "EXTRA_CFLAGS="-Wp,-D_FORTIFY_SOURCE=2 -O2"
>> causes a compiler error. It is the warning
>> 	'variable may be used uninitialized'
>> which is treated as error:
>>
>> I compile it using a FEDORA 28 installation, my gcc compiler version:
>> gcc (GCC) 8.0.1 20180324 (Red Hat 8.0.1-0.20)
>>
>> The file that causes the error is tools/lib/bpf/libbpf.c
>>
>> Here is the error message:
>>
>> [root@p23lp27] # make V=1 EXTRA_CFLAGS="-Wp,-D_FORTIFY_SOURCE=2 -O2"
>> [...]
>> Makefile.config:849: No openjdk development package found, please
>>    install JDK package, e.g. openjdk-8-jdk, java-1.8.0-openjdk-devel
>> Warning: Kernel ABI header at 'tools/include/uapi/linux/if_link.h'
>>         differs from latest version at 'include/uapi/linux/if_link.h'
>>   CC       libbpf.o
>> libbpf.c: In function âbpf_perf_event_read_simpleâ:
>> libbpf.c:2342:6: error: âretâ may be used uninitialized in this
>> 			function [-Werror=maybe-uninitialized]
>>   int ret;
>>       ^
>> cc1: all warnings being treated as errors
>> mv: cannot stat './.libbpf.o.tmp': No such file or directory
>> /home6/tmricht/linux-next/tools/build/Makefile.build:96: recipe for target 'libbpf.o' failed
>>
>> Fix this warning and add an addition check at the beginning
>> of the while loop.
>>
>> Cc: Alexei Starovoitov <ast@xxxxxxxxxx>
>> Cc: Daniel Borkmann <daniel@xxxxxxxxxxxxx>
>>
>> Suggested-by: Jakub Kicinski <jakub.kicinski@xxxxxxxxxxxxx>
>> Signed-off-by: Thomas Richter <tmricht@xxxxxxxxxxxxx>
> 
> Ah, you already sent this, LGTM, thanks Thomas!
> 
>>  tools/lib/bpf/libbpf.c | 2 ++
>>  1 file changed, 2 insertions(+)
>>
>> diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
>> index 73465caa33ba..66965ca96113 100644
>> --- a/tools/lib/bpf/libbpf.c
>> +++ b/tools/lib/bpf/libbpf.c
>> @@ -2349,6 +2349,8 @@ bpf_perf_event_read_simple(void *mem, unsigned long size,
>>  
>>  	begin = base + data_tail % size;
>>  	end = base + data_head % size;
>> +	if (begin == end)
>> +		return LIBBPF_PERF_EVENT_ERROR;
>>  
>>  	while (begin != end) {
>>  		struct perf_event_header *ehdr;

One question though, any objections to go for something like the below instead?
I doubt we ever hit this in a 'normal' situation, and given we already test for
the begin and end anyway, we could just avoid the extra test altogether. I could
change it to the below if you're good as well (no need to resend anything):

diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
index d881d37..1aafdbe 100644
--- a/tools/lib/bpf/libbpf.c
+++ b/tools/lib/bpf/libbpf.c
@@ -2273,8 +2273,8 @@ bpf_perf_event_read_simple(void *mem, unsigned long size,
        volatile struct perf_event_mmap_page *header = mem;
        __u64 data_tail = header->data_tail;
        __u64 data_head = header->data_head;
+       int ret = LIBBPF_PERF_EVENT_ERROR;
        void *base, *begin, *end;
-       int ret;

        asm volatile("" ::: "memory"); /* in real code it should be smp_rmb() */
        if (data_head == data_tail)

Thanks,
Daniel