Bug 82788 - wrong code with -fstack-clash-protection --param=stack-clash-protection-probe-interval=10 on simple code
Summary: wrong code with -fstack-clash-protection --param=stack-clash-protection-probe...
Status: RESOLVED FIXED
Alias: None
Product: gcc
Classification: Unclassified
Component: target (show other bugs)
Version: 8.0
: P3 normal
Target Milestone: ---
Assignee: Jeffrey A. Law
URL:
Keywords: wrong-code
Depends on:
Blocks:
 
Reported: 2017-10-31 19:00 UTC by Zdenek Sojka
Modified: 2017-11-06 22:45 UTC (History)
2 users (show)

See Also:
Host:
Target: x86_64-*-*, i?86-*-*
Build:
Known to work:
Known to fail: 8.0
Last reconfirmed: 2017-11-03 00:00:00

Attachments
reduced testcase (34 bytes, text/plain)
2017-10-31 19:00 UTC, Zdenek Sojka 		Details
View All Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description Zdenek Sojka 2017-10-31 19:00:23 UTC 
Created attachment 42507 [details]
reduced testcase

$ cat testcase.c 
int main() { int a[1442]; }
$ x86_64-pc-linux-gnu-gcc -fstack-clash-protection --param=stack-clash-protection-probe-interval=10 testcase.c
$ ./a.out 
Segmentation fault

In the assembly, there is a loop that never finishes (until it segfaults):
...
	lea	r11, [rsp-5120]
.LPSRL0:
	sub	rsp, 4096
	or	DWORD PTR [rsp], 0
	cmp	rsp, r11
	jne	.LPSRL0
...


$ x86_64-pc-linux-gnu-gcc -v
Using built-in specs.
COLLECT_GCC=/repo/gcc-trunk/binary-latest-amd64/bin/x86_64-pc-linux-gnu-gcc
COLLECT_LTO_WRAPPER=/repo/gcc-trunk/binary-trunk-254242-checking-yes-rtl-df-extra-nographite-amd64/bin/../libexec/gcc/x86_64-pc-linux-gnu/8.0.0/lto-wrapper
Target: x86_64-pc-linux-gnu
Configured with: /repo/gcc-trunk//configure --enable-languages=c,c++ --enable-valgrind-annotations --disable-nls --enable-checking=yes,rtl,df,extra --without-cloog --without-ppl --without-isl --build=x86_64-pc-linux-gnu --host=x86_64-pc-linux-gnu --target=x86_64-pc-linux-gnu --with-ld=/usr/bin/x86_64-pc-linux-gnu-ld --with-as=/usr/bin/x86_64-pc-linux-gnu-as --disable-libstdcxx-pch --prefix=/repo/gcc-trunk//binary-trunk-254242-checking-yes-rtl-df-extra-nographite-amd64
Thread model: posix
gcc version 8.0.0 20171030 (experimental) (GCC)
Comment 1 Jeffrey A. Law 2017-11-02 16:51:50 UTC 
Sigh.  Yet more evidence that user-configurable probing intervals is probably a bad idea...
Comment 2 Jeffrey A. Law 2017-11-06 15:51:48 UTC 
Author: law
Date: Mon Nov  6 15:51:16 2017
New Revision: 254456

URL: https://gcc.gnu.org/viewcvs?rev=254456&root=gcc&view=rev
Log:
	PR target/82788
	* config/i386/i386.c (PROBE_INTERVAL): Remove.
	(get_probe_interval): New functions.
	(ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
	(ix86_adjust_stack_and_probe): Likewise.
	(output_adjust_stack_and_probe): Likewise.
	(ix86_emit_probe_stack_range): Likewise.
	(ix86_expand_prologue): Likewise.

	PR target/82788
	* gcc.dg/pr82788.c: New test.

Added:
    trunk/gcc/testsuite/gcc.dg/pr82788.c
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/config/i386/i386.c
    trunk/gcc/testsuite/ChangeLog
Comment 3 Jeffrey A. Law 2017-11-06 22:45:51 UTC 
Fixed on the trunk this morning.