Re: [PATCH v2 3/3] perf test: Decompress kernel module before objdump
From: Adrian Hunter
Date: Mon Jun 05 2017 - 07:19:30 EST
On 03/06/17 05:00, Namhyung Kim wrote:
> If a kernel modules is compressed, it should be decompressed before
> running objdump to parse binary data correctly. This fixes a failure of
> object code reading test for me.
Makes sense to me although I have a few small comments below.
>
> Signed-off-by: Namhyung Kim <namhyung@xxxxxxxxxx>
> ---
> tools/perf/tests/code-reading.c | 26 +++++++++++++++++++++++++-
> 1 file changed, 25 insertions(+), 1 deletion(-)
>
> diff --git a/tools/perf/tests/code-reading.c b/tools/perf/tests/code-reading.c
> index 1f14e7612cbb..fe6bd34d68b1 100644
> --- a/tools/perf/tests/code-reading.c
> +++ b/tools/perf/tests/code-reading.c
> @@ -229,6 +229,7 @@ static int read_object_code(u64 addr, size_t len, u8 cpumode,
> unsigned char buf2[BUFSZ];
> size_t ret_len;
> u64 objdump_addr;
> + const char *objdump_name;
> int ret;
>
> pr_debug("Reading object code for memory address: %#"PRIx64"\n", addr);
> @@ -289,9 +290,32 @@ static int read_object_code(u64 addr, size_t len, u8 cpumode,
> state->done[state->done_cnt++] = al.map->start;
> }
>
> + objdump_name = al.map->dso->long_name;
> + if (dso__needs_decompress(al.map->dso)) {
> + char buf[32]; /* for "/tmp/perf-kmod-XXXXXX" */
We should be able to set the temporary file name buffer size without
reference to the format of the name i.e. #define it or use PATH_MAX
> +
> + if (dso__decompress_kmodule_path(al.map->dso, objdump_name,
> + buf, sizeof(buf)) < 0) {
Do we care that it is kmodule i.e. why isn't this function called just
dso__decompress_path().
> + pr_debug("decompression failed\n");
> + return -1;
> + }
> +
> + objdump_name = strdup(buf);
Why not just put the buf in the outer scope instead of using strdup().
> + if (objdump_name == NULL) {
> + pr_debug("memory allocation failed\n");
> + return -1;
> + }
> + }
> +
> /* Read the object code using objdump */
> objdump_addr = map__rip_2objdump(al.map, al.addr);
> - ret = read_via_objdump(al.map->dso->long_name, objdump_addr, buf2, len);
> + ret = read_via_objdump(objdump_name, objdump_addr, buf2, len);
> +
> + if (objdump_name != al.map->dso->long_name) {
> + unlink(objdump_name);
> + free((void *)objdump_name);
> + }
> +
> if (ret > 0) {
> /*
> * The kernel maps are inaccurate - assume objdump is right in
>