From 083488ce76ee7ab83adb51c230687a9fd67b54a7 Mon Sep 17 00:00:00 2001
From: Ken Gaillot <kgaillot@redhat.com>
Date: Fri, 16 Dec 2016 17:56:30 -0600
Subject: [PATCH] Fix: tools: avoid grep crashes in crm_report when looking for
 system logs

Grepping large binary files can exhaust memory. This avoids that and speeds up
crm_report's log scan by: skipping things that aren't files, aren't readable,
or are empty; skipping files with more than 256 nonprintable bytes in the first
1024; and using LC_ALL="C" to speed up grep.
---
 tools/report.common.in | 22 ++++++++++++++++++++--
 1 file changed, 20 insertions(+), 2 deletions(-)

diff --git a/tools/report.common.in b/tools/report.common.in
index 9b43486..70c920c 100644
--- a/tools/report.common.in
+++ b/tools/report.common.in
@@ -421,8 +421,26 @@ findmsg() {
     # Check each log file for matches.
     logfiles=""
     for f in $candidates; do
-        local cat=$(find_decompressor "$f")
-        $cat "$f" 2>/dev/null | grep -q -e "$pattern"
+        local cat=""
+
+        # We only care about readable files with something in them.
+        if [ ! -f "$f" ] || [ ! -r "$f" ] || [ ! -s "$f" ] ; then
+            continue
+        fi
+
+        cat=$(find_decompressor "$f")
+
+        # We want to avoid grepping through potentially huge binary logs such
+        # as lastlog. However, control characters sometimes find their way into
+        # text logs, so we use a heuristic of more than 256 nonprintable
+        # characters in the file's first kilobyte.
+        if [ $($cat "$f" 2>/dev/null | head -c 1024 | tr -d '[:print:][:space:]' | wc -c) -gt 256 ]
+        then
+            continue
+        fi
+
+        # Our patterns are ASCII, so we can use LC_ALL="C" to speed up grep
+        $cat "$f" 2>/dev/null | LC_ALL="C" grep -q -e "$pattern"
         if [ $? -eq 0 ]; then
 
             # Add this file to the list of hits
-- 
1.8.3.1