Re: Cleansing the environment automatically

[Greg Chicares]

On 2008-01-19 03:17Z, Paul Smith wrote:
> On Sat, 2008-01-19 at 01:09 +0000, David Wuertele wrote:
>> I always try to create a more strict configuration hygiene by
>> cleansing the environment using "env -i" at the beginning of each
>> command.
[...]
>         unexport $(.VARIABLES)

I tried that with this testcase, using a msw binary I'd built
myself from FSF make-3.81 sources, and got a segfault:

C:/tmp[0]$cat env_i.make
unexport $(.VARIABLES)

all:
        @echo xyzzy is $$xyzzy
C:/tmp[0]$make -f env_i.make
\usr\bin\make.EXE: Interrupt/Exception caught (code = 0xc0000005, addr = 
0x419fc0)

Here's a backtrace:

(gdb) r
Starting program: /usr/bin/make.exe -f env_i.make

Program received signal SIGSEGV, Segmentation fault.
0x00419fc0 in variable_hash_1 (keyv=0x429f10) at variable.c:109
109     in variable.c
(gdb) bt
#0  0x00419fc0 in variable_hash_1 (keyv=0x429f10) at variable.c:109
#1  0x0041c953 in hash_find_slot (ht=0x22f380, key=0x429f10) at hash.c:91
#2  0x0041ae09 in target_environment (file=0xbfe680) at variable.c:935
#3  0x0040d10e in start_job_command (child=0xc06fd8) at job.c:1178
#4  0x0040d38f in start_waiting_job (c=0xc06fd8) at job.c:1449
#5  0x0040dde5 in new_job (file=0xbfe680) at job.c:1728
#6  0x0041825c in update_file (file=0xbfe680, depth=0) at remake.c:1123
#7  0x00418704 in update_goal_chain (goals=0xc06dd0) at remake.c:154
#8  0x00410cc4 in main (argc=3, argv=0x3d2520, envp=0x3d29f8) at main.c:2198
#9  0x004011e7 in _end__ ()
#10 0x00401238 in mainCRTStartup ()
#11 0x77e814c7 in _libuser32_a_iname ()

I don't think I'll have time to figure this out all by myself,
but I can file a defect report on savannah if wanted.