Message-ID: <551D0507.5050107@huawei.com>
Date: Thu, 2 Apr 2015 16:59:51 +0800
From: Wang Nan <wangnan0@huawei.com>
User-Agent: Mozilla/5.0 (Windows NT 6.1; rv:24.0) Gecko/20100101 Thunderbird/24.0.1
MIME-Version: 1.0
To: Arnaldo Carvalho de Melo <acme@kernel.org>, Jiri Olsa <jolsa@redhat.com>,
        Namhyung Kim <namhyung@kernel.org>
CC: Li Zefan <lizefan@huawei.com>,
        "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>
Subject: [BUG] perf report: segfault when annotate kernel module.
Content-Type: text/plain; charset="GB2312"
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 2885
Lines: 76

Hi folks,

I hit a segfault when try to analysis a perf.data generated in an aarch 64 board.
I tried execute perf report on x86 and arm64, both of them hit similar segfault:

 $ perf report -i ./buggy/perf.data --objdump=/path/to/aarch64-oe-linux-objdump --kallsyms /path/to/kallsyms

perf: Segmentation fault
-------- backtrace --------
/home/user/perf[0x503478]
/lib64/libc.so.6(+0x3545f)[0x7f5e6944b45f]
/home/user/perf[0x499b56]
/home/user/perf(dso__load_kallsyms+0x13c)[0x49b56c]
/home/user/perf(dso__load+0x72e)[0x49c21e]
/home/user/perf(map__load+0x6e)[0x4ae9ee]
/home/user/perf(thread__find_addr_map+0x24c)[0x47deec]
/home/user/perf(perf_event__preprocess_sample+0x88)[0x47e238]
/home/user/perf[0x43ad02]
/home/user/perf[0x4b55bc]
/home/user/perf(ordered_events__flush+0x1a9)[0x4b58c9]
/home/user/perf(perf_session__process_events+0x407)[0x4b42d7]
/home/user/perf(cmd_report+0xf11)[0x43bfc1]
/home/user/perf[0x474702]
/home/user/perf(main+0x5f5)[0x42de95]
/lib64/libc.so.6(__libc_start_main+0xf4)[0x7f5e69437bd4]
/home/user/perf[0x42dfc4]

I did some debugging and found the problem is raised at dso__load_kcore:

static int dso__load_kcore(struct dso *dso, struct map *map,
                           const char *kallsyms_filename)
{
        struct map_groups *kmaps = map__kmap(map)->kmaps;    // kmap is NULL here
        struct machine *machine = kmaps->machine;            // segfault.
        ...
}

Then I realised that if a struct map is created by machine__new_module with map__new2,
the map__kmap(map) is never set. However, if I set

        map->map_ip = map->unmap_ip = identity__map_ip;
        kmap = map__kmap(map);
        kmap->kmaps = &machine->kmaps;

in machine__new_module like in __machine__create_kernel_maps, I hit another problem:

*** Error in `/home/user/perf': free(): invalid next size (normal): 0x0000000002411a60 ***
======= Backtrace: =========
/lib64/libc.so.6(+0x6eeef)[0x7f2ed502feef]
/lib64/libc.so.6(+0x78cae)[0x7f2ed5039cae]
/lib64/libc.so.6(+0x79987)[0x7f2ed503a987]
/home/user/perf(dso__data_fd+0x456)[0x496be6]
/home/user/perf(dso__type+0x1a)[0x4987ea]
/home/user/perf(vdso__dso_findnew+0x7d)[0x4d2cbd]
/home/user/perf(map__new+0x2c3)[0x4ae7c3]
/home/user/perf[0x4a90fa]
/home/user/perf(machine__process_mmap2_event+0xc1)[0x4acbf1]
/home/user/perf[0x4b1fec]
/home/user/perf(perf_session__process_events+0x3bf)[0x4b45ff]
/home/user/perf(cmd_report+0xf12)[0x43bfc2]
/home/user/perf[0x474703]
/home/user/perf(main+0x5f6)[0x42de96]
/lib64/libc.so.6(__libc_start_main+0xf5)[0x7f2ed4fe2bd5]
/home/user/perf[0x42dfc5]
======= Memory map: ========
...

Does anyone has any idea on it?

Thank you!

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/