From 017e6c97955c652bc4a18fda949a4ed291406cbe Mon Sep 17 00:00:00 2001
From: Ben Vandervalk <benv@bcgsc.ca>
Date: Tue, 4 Dec 2018 10:39:37 -0800
Subject: [PATCH] abyss-pe: increase stack size limit for certain ABySS
 programs

We were seeing stack overflows in the `abyss-bloom-dbg` program on conifer
genomes, caused by graph searches implemented with recursion. We also
occasionally see stack overflows in `SimpleGraph` and `PathConsensus`, also due
to recursive graph searches.

As a workaround for this problem, I have added a wrapper script called
`abyss-stack-size` which runs a shell command with an increased max stack size,
using `ulimit -s`.
---
 bin/Makefile.am      |  4 +++-
 bin/abyss-pe         | 13 ++++++++-----
 bin/abyss-stack-size | 25 +++++++++++++++++++++++++
 3 files changed, 36 insertions(+), 6 deletions(-)
 create mode 100755 bin/abyss-stack-size

diff --git a/bin/Makefile.am b/bin/Makefile.am
index 624de4134..8563665f3 100644
--- a/bin/Makefile.am
+++ b/bin/Makefile.am
@@ -17,8 +17,10 @@ dist_bin_SCRIPTS = \
 	abyss-samtoafg \
 	abyss-tabtomd \
 	abyss-bloom-dist.mk \
-	abyss-dida
+	abyss-dida \
+	abyss-stack-size \
 	$(nompi)
+
 dist_noinst_SCRIPTS = \
 	abyss-adjtodot.pl \
 	abyss-cstont \
diff --git a/bin/abyss-pe b/bin/abyss-pe
index 663f425be..2bbbfdc98 100755
--- a/bin/abyss-pe
+++ b/bin/abyss-pe
@@ -21,6 +21,9 @@ gtime=command time -v -o $@.time
 endif
 endif
 
+# Wrapper script for commands that require an increased stack size limit
+stack=abyss-stack-size 65536
+
 # Define this environment variable on Mac OS X to read
 # compressed files.
 export DYLD_FORCE_FLAT_NAMESPACE=1
@@ -532,7 +535,7 @@ endif
 
 ifdef B
 %-1.fa:
-	$(gtime) abyss-bloom-dbg $(abyssopt) $(ABYSS_OPTIONS) $(in) $(se) > $@
+	$(gtime) $(stack) abyss-bloom-dbg $(abyssopt) $(ABYSS_OPTIONS) $(in) $(se) > $@
 else ifdef K
 
 ifdef np
@@ -626,7 +629,7 @@ endif
 # Assemble contigs
 
 %-4.path1: %-4.$g %-3.dist
-	$(gtime) SimpleGraph $v $(sgopt) $(SIMPLEGRAPH_OPTIONS) -j$j -k$k -o $@ $^
+	$(gtime) $(stack) SimpleGraph $v $(sgopt) $(SIMPLEGRAPH_OPTIONS) -j$j -k$k -o $@ $^
 
 %-4.path2: %-4.path1 %-3.fa.fai %-4.fa.fai
 	cat $*-3.fa.fai $*-4.fa.fai \
@@ -639,7 +642,7 @@ ifndef cs
 
 %-5.path %-5.fa %-5.$g: %-3.fa %-4.fa %-4.$g %-4.path3
 	cat $(wordlist 1, 2, $^) \
-		|$(gtime) PathConsensus $v --$g -k$k $(pcopt) $(PATHCONSENSUS_OPTIONS) -o $*-5.path -s $*-5.fa -g $*-5.$g - $(wordlist 3, 4, $^)
+		|$(gtime) $(stack) PathConsensus $v --$g -k$k $(pcopt) $(PATHCONSENSUS_OPTIONS) -o $*-5.path -s $*-5.fa -g $*-5.$g - $(wordlist 3, 4, $^)
 
 %-6.fa: %-3.fa %-4.fa %-5.fa %-5.$g %-5.path
 	cat $(wordlist 1, 3, $^) |$(gtime) MergeContigs $(mcopt) -o $@ - $(wordlist 4, 5, $^)
@@ -704,7 +707,7 @@ endif
 	$(gtime) abyss-scaffold $(scopt) -s$S -n$N -g $@.dot $(SCAFFOLD_OPTIONS) $^ >$@
 
 %-7.path %-7.$g %-7.fa: %-6.fa %-6.$g %-6.path
-	$(gtime) PathConsensus $v --$g -k$k $(pcopt) $(PATHCONSENSUS_OPTIONS) -s $*-7.fa -g $*-7.$g -o $*-7.path $^
+	$(gtime) $(stack) PathConsensus $v --$g -k$k $(pcopt) $(PATHCONSENSUS_OPTIONS) -s $*-7.fa -g $*-7.$g -o $*-7.path $^
 
 %-8.fa: %-6.fa %-7.fa %-7.$g %-7.path
 	cat $(wordlist 1, 2, $^) \
@@ -850,7 +853,7 @@ sealer_ks?=-k90 -k80 -k70 -k60 -k50 -k40 -k30
 	$(gtime) abyss-scaffold $(scopt) -s$S -n1 -g $@.$g $(SCAFFOLD_OPTIONS) $^ >$@
 
 %-9.path %-9.$g %-9.fa: %-8.fa %-8.$g %-8.path
-	$(gtime) PathConsensus $v --$g -k$k $(pcopt) $(PATHCONSENSUS_OPTIONS) -s $*-9.fa -g $*-9.$g -o $*-9.path $^
+	$(gtime) $(stack) PathConsensus $v --$g -k$k $(pcopt) $(PATHCONSENSUS_OPTIONS) -s $*-9.fa -g $*-9.$g -o $*-9.path $^
 
 %-10.fa: %-8.fa %-9.fa %-9.$g %-9.path
 	cat $(wordlist 1, 2, $^) \
diff --git a/bin/abyss-stack-size b/bin/abyss-stack-size
new file mode 100755
index 000000000..8546ad09f
--- /dev/null
+++ b/bin/abyss-stack-size
@@ -0,0 +1,25 @@
+#!/bin/sh
+
+if [ $# -lt 2 ]; then
+	echo "Usage: $(basename $0) <STACK_SIZE> <COMMAND>" >&2
+	echo "Run COMMAND in a shell with a maximum stack size of" >&2
+	echo "at least STACK_SIZE in kilobytes." >&2
+	exit 1
+fi
+min_stack=$1; shift
+
+# Note: A max stack size of "unlimited" may not actually
+# be unlimited. For example, on Linux, using "unlimited"
+# results in a max stack size of 2 MB, which
+# is less than the default max stack size of 8 MB.
+# How confusing!
+
+stack=$(ulimit -s)
+if [ "$stack" = "unlimited" ] || [ "$stack" -lt "$min_stack" ]; then
+	ulimit -s $min_stack
+fi
+stack=$(ulimit -s)
+
+echo "Running with max stack size of $stack KB: $*" >&2
+exec /bin/sh -c "$*"
+