[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <08c6ad07-7994-3e56-b20e-d75727ca7765@intel.com>
Date: Tue, 28 Feb 2017 12:02:36 +0200
From: Adrian Hunter <adrian.hunter@...el.com>
To: Arnaldo Carvalho de Melo <acme@...nel.org>
Cc: Jiri Olsa <jolsa@...nel.org>, Namhyung Kim <namhyung@...nel.org>,
David Ahern <dsahern@...il.com>,
Linux Kernel Mailing List <linux-kernel@...r.kernel.org>
Subject: Re: perf test 'instruction decoder' failing
On 24/02/17 22:27, Arnaldo Carvalho de Melo wrote:
> Hi Adrian,
>
> This test is failing, I'll try to recheck this and bisect it,
> but since I haven't had the time so far and I saw it since yesterday (at
> least), lemme document it here:
>
>
> [root@...et ~]# perf test -v 57
> 57: x86 instruction decoder - new instructions :
> --- start ---
> test child forked, pid 18975
> Decoded ok: 0f 31 rdtsc
> Decoded ok: 0f 1b 80 78 56 34 12 bndstx %bnd0,0x12345678(%eax)
> <SNIP tons of OKays>
> Decoded ok: 0f 1b 85 78 56 34 12 bndstx %bnd0,0x12345678(%ebp)
> Decoded ok: 0f 1b 84 01 78 56 34 12 bndstx %bnd0,0x12345678(%ecx,%eax,1)
> Decoded ok: 0f 1b 84 05 78 56 34 12 bndstx %bnd0,0x12345678(%ebp,%eax,1)
> Decoded ok: 0f 1b 84 08 78 56 34 12 bndstx %bnd0,0x12345678(%eax,%ecx,1)
> Decoded ok: f2 e8 fc ff ff ff bnd call fce <main+0xfce>
> Failed to decode 'rel' value (0xfffffffc vs expected 0): f2 ff 10 bnd call *(%eax)
> Failed to decode 'rel' value (0xfffffffc vs expected 0): f2 c3 bnd ret
> Decoded ok: f2 e9 fc ff ff ff bnd jmp fd9 <main+0xfd9>
> Decoded ok: f2 e9 fc ff ff ff bnd jmp fdf <main+0xfdf>
> Failed to decode 'rel' value (0xfffffffc vs expected 0): f2 ff 21 bnd jmp *(%ecx)
> Decoded ok: f2 0f 85 fc ff ff ff bnd jne fe9 <main+0xfe9>
> Failed to decode 'rel' value (0xfffffffc vs expected 0): 0f 3a cc c1 00 sha1rnds4 $0x0,%xmm1,%xmm0
> Failed to decode 'rel' value (0xfffffffc vs expected 0): 0f 3a cc d7 91 sha1rnds4 $0x91,%xmm7,%xmm2
> Failed to decode 'rel' value (0xfffffffc vs expected 0): 0f 3a cc 00 91 sha1rnds4 $0x91,(%eax),%xmm0
> Failed to decode 'rel' value (0xfffffffc vs expected 0): 0f 3a cc 05 78 56 34 12 91 sha1rnds4 $0x91,0x12345678,%xmm0
> Failed to decode 'rel' value (0xfffffffc vs expected 0): 0f 3a cc 18 91 sha1rnds4 $0x91,(%eax),%xmm3
> Failed to decode 'rel' value (0xfffffffc vs expected 0): 0f 3a cc 04 01 91 sha1rnds4 $0x91,(%ecx,%eax,1),%xmm0
> Failed to decode 'rel' value (0xfffffffc vs expected 0): 0f 3a cc 04 05 78 56 34 12 91 sha1rnds4 $0x91,0x12345678(,%eax,1),%xmm0
> Failed to decode 'rel' value (0xfffffffc vs expected 0): 0f 3a cc 04 08 91 sha1rnds4 $0x91,(%eax,%ecx,1),%xmm0
> Failed to decode 'rel' value (0xfffffffc vs expected 0): 0f 3a cc 04 c8 91 sha1rnds4 $0x91,(%eax,%ecx,8),%xmm0
> <SNIP lots of fails>
> Failed to decode 'rel' value (0xfffffffc vs expected 0): 0f 1b 45 12 bndstx %bnd0,0x12(%rbp)
> Failed to decode 'rel' value (0xfffffffc vs expected 0): 0f 1b 44 01 12 bndstx %bnd0,0x12(%rcx,%rax,1)
> Failed to decode 'rel' value (0xfffffffc vs expected 0): 0f 1b 44 05 12 bndstx %bnd0,0x12(%rbp,%rax,1)
> Failed to decode 'rel' value (0xfffffffc vs expected 0): 0f 1b 44 08 12 bndstx %bnd0,0x12(%rax,%rcx,1)
> Failed to decode 'rel' value (0xfffffffc vs expected 0): 0f 1b 80 78 56 34 12 bndstx %bnd0,0x12345678(%rax)
> Failed to decode 'rel' value (0xfffffffc vs expected 0): 0f 1b 85 78 56 34 12 bndstx %bnd0,0x12345678(%rbp)
> Failed to decode 'rel' value (0xfffffffc vs expected 0): 0f 1b 84 01 78 56 34 12 bndstx %bnd0,0x12345678(%rcx,%rax,1)
> Failed to decode 'rel' value (0xfffffffc vs expected 0): 0f 1b 84 05 78 56 34 12 bndstx %bnd0,0x12345678(%rbp,%rax,1)
> Failed to decode 'rel' value (0xfffffffc vs expected 0): 0f 1b 84 08 78 56 34 12 bndstx %bnd0,0x12345678(%rax,%rcx,1)
> Decoded ok: f2 e8 00 00 00 00 bnd callq f22 <main+0xf22>
> Decoded ok: 67 f2 ff 10 bnd callq *(%eax)
> <SNIP lots of OKays>
> Decoded ok: 41 0f c7 9c c8 78 56 34 12 xrstors 0x12345678(%r8,%rcx,8)
> Decoded ok: 0f 01 ee rdpkru
> Decoded ok: 0f 01 ef wrpkru
> test child finished with -1
> ---- end ----
> x86 instruction decoder - new instructions: FAILED!
>
> Fedora 25,
>
> [acme@...et linux]$ uname -a
> Linux jouet 4.10.0-rc8 #2 SMP Wed Feb 15 15:26:36 BRT 2017 x86_64 x86_64 x86_64 GNU/Linux
>
> [acme@...et linux]$ rpm -q binutils elfutils gcc glibc
> binutils-2.26.1-1.fc25.x86_64
> elfutils-0.168-1.fc25.x86_64
> gcc-6.3.1-1.fc25.x86_64
> glibc-2.24-4.fc25.x86_64
> glibc-2.24-4.fc25.i686
> [acme@...et linux]$
>
> [acme@...et linux]$ perf -v
> perf version 4.10.g264b77
>
> perf as in tip/perf/core and soon as in acme/perf/core
There is missing initialization. It only affects the test because it is checking 'rel' even in cases where there is no value.
Here is the fix.
diff --git a/tools/perf/util/intel-pt-decoder/intel-pt-insn-decoder.c b/tools/perf/util/intel-pt-decoder/intel-pt-insn-decoder.c
index 7913363bde5c..55b6250350d7 100644
--- a/tools/perf/util/intel-pt-decoder/intel-pt-insn-decoder.c
+++ b/tools/perf/util/intel-pt-decoder/intel-pt-insn-decoder.c
@@ -39,6 +39,8 @@ static void intel_pt_insn_decoder(struct insn *insn,
enum intel_pt_insn_branch branch = INTEL_PT_BR_NO_BRANCH;
int ext;
+ intel_pt_insn->rel = 0;
+
if (insn_is_avx(insn)) {
intel_pt_insn->op = INTEL_PT_OP_OTHER;
intel_pt_insn->branch = INTEL_PT_BR_NO_BRANCH;
Powered by blists - more mailing lists