Re: results of gnulib tests with -fsanitize=address

[Bruno Haible]

Paul Eggert wrote:
> I installed the attached patch to work around the problem.

Thanks! I confirm that 'test-readtokens.sh' now properly reports a memory leak:

FAIL: test-readtokens.sh
========================


=================================================================
==22965==ERROR: LeakSanitizer: detected memory leaks

Direct leak of 128 byte(s) in 1 object(s) allocated from:
    #0 0x7f3d77fa7961 in realloc 
(/usr/lib/x86_64-linux-gnu/libasan.so.2+0x98961)
    #1 0x40632c in xrealloc ../../gllib/xmalloc.c:61

SUMMARY: AddressSanitizer: 128 byte(s) leaked in 1 allocation(s).
FAIL test-readtokens.sh (exit status: 1)


> Similarly, 
> we should advise people not to run 'configure' with sanitization options 
> like memory-leak detection that affect valid (albeit inefficient) test 
> programs.
> 
> However, it should be OK to configure with -fsanitize=address if the 
> ASAN_OPTIONS environment variable suppresses checks for 
> valid-albeit-inefficent behavior. For example, 
> ASAN_OPTIONS='detect_leaks=0' may suffice. (I haven't checked this.)

Good to hear this, thanks. So, people should
  1. set ASAN_OPTIONS='detect_leaks=0' [1]
  2. run configure
  3. set ASAN_OPTIONS to whatever value they need [2]
  4. run "make check"

Bruno

[1] https://github.com/google/sanitizers/wiki/AddressSanitizerLeakSanitizer
[2] https://github.com/google/sanitizers/wiki/AddressSanitizerFlags