diff options
Diffstat (limited to 'results/classifier/zero-shot/108/other/1661815')
| -rw-r--r-- | results/classifier/zero-shot/108/other/1661815 | 58 |
1 files changed, 58 insertions, 0 deletions
diff --git a/results/classifier/zero-shot/108/other/1661815 b/results/classifier/zero-shot/108/other/1661815 new file mode 100644 index 000000000..5668d1d14 --- /dev/null +++ b/results/classifier/zero-shot/108/other/1661815 @@ -0,0 +1,58 @@ +vnc: 0.749 +graphic: 0.734 +performance: 0.728 +device: 0.725 +other: 0.708 +network: 0.677 +permissions: 0.661 +debug: 0.656 +PID: 0.623 +files: 0.617 +socket: 0.557 +semantic: 0.542 +KVM: 0.451 +boot: 0.368 + +Stack address is returned from function translate_one + +The vulnerable version is qemu-2.8.0, and the vulnerable function is in "target-s390x/translate.c". + +The code snippet is as following. + +static ExitStatus translate_one(CPUS390XState *env, DisasContext *s) +{ + const DisasInsn *insn; + ExitStatus ret = NO_EXIT; + DisasFields f; + ... + s->fields = &f; + ... + s->pc = s->next_pc; + return ret; +} + +A stack address, i.e. the address of local variable "f" is returned from current function through the output parameter "s->fields" as a side effect. + +This issue is one kind of undefined behaviors, according the C Standard, 6.2.4 [ISO/IEC 9899:2011] (https://www.securecoding.cert.org/confluence/display/c/DCL30-C.+Declare+objects+with+appropriate+storage+durations) + +This dangerous defect may lead to an exploitable vulnerability. +We suggest sanitizing "s->fields" as null before return. + +Note that this issue is reported by shqking and Zhenwei Zou together. + +The calling function never uses "->fields", so I do not see a real vulnerability here, is there? Did you use a code analyser for this, or how did you come across this issue? + +Thanks for your reply. + +Inspired by this issue in apache httpd (https://bz.apache.org/bugzilla/show_bug.cgi?id=59844#c0), +we customized a checker based on the Clang Static Analyzer to detect such undefined behavior. + +Yes. +After examining the code carefully, we didn't find any place where the "->fields" is accessed, either. However, we think this kind of defect seems like a 'time bomb' and we'd better fix it just to be on the safe side. + +I've finally posted a patch for this: +https://lists.gnu.org/archive/html/qemu-devel/2020-01/msg05204.html + +Fixed here: +https://git.qemu.org/?p=qemu.git;a=commitdiff;h=344a7f656e8d211cdd6e + |