Package: conky
Version: 1.10.6-1.1bds
Severity: important
Tags: patch
This patch corrects a segfault (and general instability) associated with
CPU monitoring $cpugraph, $cpubar, and related load monitoring
directives that occurred when the kernel reports a non-sequential list
of CPU indices on an SMP system.
This is a new bug that arrived with the 'stretch' version of conky.
For example, "/sys/devices/system/cpu/present" for my AMD FX(tm)-6350
Six-Core Processor reports "0,3-7" (stock debian kernel 4.9.0-3-amd64 ).
I assume that chip is really an 8-core die with two cores disabled...
Presumably you could also get "0,3-4,6", and other combos too.
This bug and patch has also been reported upstream:
https://github.com/brndnmtthws/conky/pull/416
This may be the same problem reported in Bug #863144: "conky: seqfaults"
-- Brad
-- System Information:
Debian Release: 9.1
APT prefers stable
APT policy: (900, 'stable'), (80, 'unstable'), (60, 'experimental')
Architecture: amd64 (x86_64)
Foreign Architectures: i386
Kernel: Linux 4.9.0-3-amd64 (SMP w/6 CPU cores)
Locale: LANG=en_US.UTF-8, LC_CTYPE=en_US.UTF-8 (charmap=UTF-8),
LANGUAGE=en_US.UTF-8 (charmap=UTF-8)
Shell: /bin/sh linked to /bin/bash
Init: systemd (via /run/systemd/system)
Versions of packages conky depends on:
ii conky-all 1.10.6-1
conky recommends no packages.
conky suggests no packages.
-- no debconf information
From d3065426868cdf63ccbbaf6afd917de64403ae49 Mon Sep 17 00:00:00 2001
From: Brad Sawatzky <brad+deb...@swatter.net>
Date: Sun, 27 Aug 2017 16:10:48 -0400
Subject: [PATCH] Fix segfault when kernel reports non-sequential SMP CPU
indices
- For an AMD FX(tm)-6350 Six-Core Processor the file '/sys/.../present'
reports "0,3-7". I assume that chip is really an 8-core die with two
cores disabled... Presumably you could also get "0,3-4,6", and other
combos too...
---
src/linux.cc | 32 +++++++++++++++++++++++++++-----
1 file changed, 27 insertions(+), 5 deletions(-)
diff --git a/src/linux.cc b/src/linux.cc
index 664ddf5e..5ebed315 100644
--- a/src/linux.cc
+++ b/src/linux.cc
@@ -801,8 +801,11 @@ void get_cpu_count(void)
{
FILE *stat_fp;
static int rep = 0;
- int highest_cpu_index;
char buf[256];
+ char *str1, *str2, *token, *subtoken;
+ char *saveptr1, *saveptr2;
+ int subtoken1=-1;
+ int subtoken2=-1;
if (info.cpu_usage) {
return;
@@ -819,11 +822,30 @@ void get_cpu_count(void)
break;
}
- if (sscanf(buf, "%*d-%d", &highest_cpu_index) == 1) {
- info.cpu_count = highest_cpu_index;
+ // Do some parsing here to handle skipped cpu numbers. For example,
+ // for an AMD FX(tm)-6350 Six-Core Processor /sys/.../present reports
+ // "0,3-7". I assume that chip is really an 8-core die with two cores
+ // disabled... Presumably you could also get "0,3-4,6", and other
+ // combos too...
+ for (str1 = buf; ; str1 = NULL) {
+ token = strtok_r(str1, ",", &saveptr1);
+ if (token == NULL) break;
+ ++info.cpu_count;
+
+ subtoken1=-1;
+ subtoken2=-1;
+ for (str2 = token; ; str2 = NULL) {
+ subtoken = strtok_r(str2, "-", &saveptr2);
+ if (subtoken == NULL) break;
+ if(subtoken1 < 0)
+ subtoken1=atoi(subtoken);
+ else
+ subtoken2=atoi(subtoken);
+ }
+ if(subtoken2 > 0)
+ info.cpu_count += subtoken2 - subtoken1;
}
}
- ++info.cpu_count;
info.cpu_usage = (float*)malloc((info.cpu_count + 1) * sizeof(float));
fclose(stat_fp);
@@ -896,7 +918,7 @@ int update_stat(void)
} else if (strncmp(buf, "cpu", 3) == 0) {
double delta;
if (isdigit(buf[3])) {
- idx = atoi(&buf[3]) + 1;
+ idx++; //just increment here since the CPU index can skip numbers
} else {
idx = 0;
}
--
2.11.0
