add microcode version to the hs_err files

[Ivanov, Vladimir A]

>  +#if defined(IA32) || defined(AMD64)
>
> Is that not synonymous with x86?
This patter was copied from the method “print_model_name_and_flags” (file os/linux/os_linux.cpp).
This method also read the “/proc/cpuinfo” file and I reuse it as ‘template’ for the new method.
It is better to use one pattern to work with exactly same file but in general you are right.
The X86 is defined in the file ./share/utilities/macros.hpp as:
#if defined(IA32) || defined(AMD64)
#define X86
#define X86_ONLY(code) code
#define NOT_X86(code)

The question here: could I delete this “ifdefs” while this method should work on x86 only?

Thanks, Vladimir

From: Thomas Stüfe <thomas.stuefe at gmail.com>
Sent: Friday, July 17, 2020 2:26 PM
To: Ivanov, Vladimir A <vladimir.a.ivanov at intel.com>; Hotspot dev runtime <hotspot-runtime-dev at openjdk.java.net>
Cc: hotspot-compiler-dev at openjdk.java.net
Subject: Re: add microcode version to the hs_err files



On Fri, Jul 17, 2020 at 11:19 PM Thomas Stüfe <thomas.stuefe at gmail.com<mailto:thomas.stuefe at gmail.com>> wrote:
Hi Vladimir,

I think this would be more suited to hotspot-runtime.

http://cr.openjdk.java.net/~sviswanathan/Vladimir/8249672/webrev.00/src/hotspot/os_cpu/linux_x86/os_linux_x86.cpp.udiff.html

+#if defined(IA32) || defined(AMD64)

Is that not synonymous with x86?

+    while ((read = getline(&line, &len, fp)) != -1) {
+      if (len > 10 && strstr(line, "microcode") != NULL) {
+        char* rev = strchr(line, ':');
+        if (rev != NULL) sscanf(rev + 1, "%x", &result);
+        break;
+      }
+    }
+    free(line);

Not sure this works as intended. At the first call to getline() it will allocate a line buffer for you and return it. That buffer will be as large as the first line you happen to read. You then pass that same buffer into getline to fetch the next lines, but what if those are longer than the first?


Forget that point, getline calls realloc() on the line buffer to resize it, so this should be okay.

Thanks, Thomas

But anyway it would be better to pass a simple caller provided buffer in - stack allocated. Since this function is called at crash time and the C heap could be corrupted.

Cheers, Thomas


On Fri, Jul 17, 2020 at 10:22 PM Ivanov, Vladimir A <vladimir.a.ivanov at intel.com<mailto:vladimir.a.ivanov at intel.com>> wrote:
Hello,

could you please review the patch  http://cr.openjdk.java.net/~sviswanathan/Vladimir/8249672/webrev.00/

This patch add the microcode version for different OSes that may be useful in the issue resolution process.



The reported microcode version for different OSes loos as:



Linux (RHEL7.7):

# cat hs_err_pid251046.log |grep microc

CPU: total 112 (initial active 112) (28 cores per cpu, 2 threads per core) family 6 model 85 stepping 4 microcode 0x200005e, cmov, cx8, fxsr, mmx, sse, sse2, sse3, ssse3, sse4.1, sse4.2, popcnt, vzeroupper, avx, avx2, aes, clmul, erms, rtm, 3dnowpref, lzcnt, ht, tsc, tscinvbit, bmi1, bmi2, adx, fma, clflush, clflushopt, clwb



Windows (Win10, v1809):

CPU: total 4 (initial active 4) (2 cores per cpu, 2 threads per core) family 6 model 142 stepping 9 microcode 0xb4, cmov, cx8, fxsr, mmx, sse, sse2, sse3, ssse3, sse4.1, sse4.2, popcnt, vzeroupper, avx, avx2, aes, clmul, erms, rtm, 3dnowpref, lzcnt, ht, tsc, tscinvbit, bmi1, bmi2, adx, fma, clflush, clflushopt



MacOS (Darwin):

$ cat hs_err_pid95187.log |grep microc

CPU: total 8 (initial active 8) (4 cores per cpu, 2 threads per core) family 6 model 126 stepping 5 microcode 0x78, cmov, cx8, fxsr, mmx, sse, sse2, sse3, ssse3, sse4.1, sse4.2, popcnt, vzeroupper, avx, avx2, aes, clmul, erms, 3dnowpref, lzcnt, ht, tsc, tscinvbit, bmi1, bmi2, adx, sha, fma, clflush, clflushopt



Thanks, Vladimir


  Thanks, Vladimir