[prev in list] [next in list] [prev in thread] [next in thread] 

List:       pcc-list
Subject:    Function arguments related segfault on amd64
From:       Henri Kemppainen <duclare () guu ! fi>
Date:       2011-07-20 11:23:02
Message-ID: 87livtdwmh.fsf () guu ! fi
[Download RAW message or body]

% cat fuu.c
void fun(int badarg[][4][4]) {
        return;
}

% ~/pre/bin/pcc -v fuu.c
pcc 1.1.0.DEVEL 20110719 for x86_64-unknown-openbsd4.9
/home/duclare/pre/libexec/cpp -v -D__PCC__=1 -D__PCC_MINOR__=1
-D__PCC_MINORMINOR__=0 -D__GNUC__=4 -D__GNUC_MINOR__=3
-D__GNUC_PATCHLEVEL__=1 -D__GNUC_STDC_INLINE__=1 -D__VERSION__="pcc
1.1.0.DEVEL 20110719 for x86_64-unknown-openbsd4.9" -D__SCHAR_MAX__=127
-D__SHRT_MAX__=32767 -D__INT_MAX__=0x7fffffff
-D__LONG_MAX__=0x7fffffffffffffffL
-D__LONG_LONG_MAX__=0x7fffffffffffffffLL -D__REGISTER_PREFIX__=
-D__USER_LABEL_PREFIX__= -D__OpenBSD__ -D__ELF__
-D__STDC_ISO_10646__=200009L -D__WCHAR_TYPE__=int -D__SIZEOF_WCHAR_T__=4
-D__WCHAR_MAX__=2147483647 -D__WINT_TYPE__=unsigned int
-D__SIZE_TYPE__=unsigned long -D__PTRDIFF_TYPE__=long int
-D__SIZEOF_WINT_T__=4 -D__amd64__ -S /home/duclare/pre/include/pcc/ -S
/usr/include/ -S
/home/duclare/pre/lib/pcc/x86_64-unknown-openbsd4.9/1.1.0.DEVEL/include/
fuu.c /tmp/ctm.EJhM98 
cpp: pcc 1.1.0.DEVEL 20110719 for x86_64-unknown-openbsd4.9
/home/duclare/pre/libexec/ccom -v /tmp/ctm.EJhM98 /tmp/ctm.2nTebM 
ccom: pcc 1.1.0.DEVEL 20110719 for x86_64-unknown-openbsd4.9
major internal compiler error: fuu.c, line 1

% gdb ~/pre/libexec/ccom
(gdb) run fuu.c
Starting program: /home/duclare/pre/libexec/ccom fuu.c

Program received signal SIGSEGV, Segmentation fault.
0x0000000000409f4d in bfcode (s=0x20b8df130, cnt=1)
    at ../../arch/amd64/code.c:309
309             for (; al->type != TELLIPSIS; al++) {
(gdb) bt
#0  0x0000000000409f4d in bfcode (s=0x20b8df130, cnt=1)
    at ../../arch/amd64/code.c:309
#1  0x0000000000420d4c in dclargs () at ./pftn.c:640
#2  0x00000000004073bb in yyparse () at cgram.y:771
#3  0x00000000004161c0 in main (argc=1, argv=0x7f7ffffbd000) at
./main.c:324

It looks like this loop in bfcode() skips over the TNULL that should
terminate it.  I didn't dig deep enough to figure out if it's a bug
in the loop or somewhere up the stream.  Didn't test on i386.

[prev in list] [next in list] [prev in thread] [next in thread]