Several heap buffer overflow issues have been found. #189
Comments
|
When I run this through mkd2html on my dev machine, I don't get any warnings on the internal debugging malloc that I put into discount (except for the two memory blocks that contain the source filename & output html name, which are expected) so I've a couple of questions:
|
|
Sorry for replying so late. |
Some other issues have been found by fuzzing process, the executable file is mkd2html, there are totally 2 issues.
|
|
Once you pointed me at the offending compiler flag I was able to reproduce it (and holy cow the code slows waaaay down when built with -fsanitize=address -- the clang documentation that says it doesn't slow things down much is a baldfaced lie) and fix it (normally I null-terminate input lines by doing an (And I also found a bug where I was setting the dle of the new line from the dle of the old line, so a bit of code like {3 spaces at the start of the line} ANYWAY, I've pushed a little patch that makes -fsanitize=address NOT abend on your issue2_testcase (and I suspect not abend on your issue1 & issue3 testcases, because the offending data block is allocated in |
|
The issues from #189 (comment) have been assigned CVE-2018-11503 and CVE-2018-11504 respectively. |
Another issue has been found when I fuzz markdown.Details are displayed as follows:
|
|
How are you calling the markdown program? |
|
I just call markdown without any parameter : ./markdown input, and the input is the test case I record at https://github.com/fCorleone/fuzz_programs/blob/master/discount/issue4_testcase. |
|
So you're doing by itself. What compiler flags are you passing, and what configure.sh options are you using? I couldn't repeat the fault here, and this is configuring discount as |
|
I use afl-gcc as the compiler when I run the script configure. The flags are the same. |
|
Do you do any of the configure.sh options, or just ? (I'm on a mac, so no afl-gcc for me; clang's sanitize=address is good for catching boundary overflows) |
|
just clang may work too. |
|
Clang on macos ( Can you bisect the input file to see if you can reproduce the fault with a shorter test case? If I can't duplicate the error on my build machines I'll have to walk the code by eye to see if I can figure out the problem, and I'd much rather do that with a considerably shorter piece of source. |
|
And also which version of discount are you using; top of source or an earlier version? |
|
I'm going to close this issue because all of your overflow catches involved splitline(), which I reworked to null-terminate the lines after splitting. If you find anything that makes afl freak out on top-of-source, let me know and I'll reopen it. |
Hi, Can you refer me to the commit that fix this issue ? |
A buffer overflow has been discovered. The information is displayed as follows:
==4515==ERROR: AddressSanitizer: heap-buffer-overflow on address 0x60b00000a91a at pc 0x7f7e7a5e7df8 bp 0x7ffe878efa50 sp 0x7ffe878ef1f8
READ of size 103 at 0x60b00000a91a thread T0
#0 0x7f7e7a5e7df7 in __asan_memmove (/usr/lib/x86_64-linux-gnu/libasan.so.2+0x8cdf7)
#1 0x40b516 in memmove /usr/include/x86_64-linux-gnu/bits/string3.h:59
#2 0x40b516 in __mkd_trim_line /home/ubuntu/mfc_fuzz/discount/mkdio.c:85
#3 0x4260e0 in codeblock /home/ubuntu/mfc_fuzz/discount/markdown.c:611
#4 0x4260e0 in compile /home/ubuntu/mfc_fuzz/discount/markdown.c:1317
#5 0x42a56e in compile_document /home/ubuntu/mfc_fuzz/discount/markdown.c:1230
#6 0x42a56e in mkd_compile /home/ubuntu/mfc_fuzz/discount/markdown.c:1436
#7 0x40272c in main /home/ubuntu/mfc_fuzz/discount/mkd2html.c:177
#8 0x7f7e7a1b182f in __libc_start_main (/lib/x86_64-linux-gnu/libc.so.6+0x2082f)
#9 0x403d48 in _start (/home/ubuntu/mfc_fuzz/discount/mkd2html+0x403d48)
0x60b00000a91a is located 0 bytes to the right of 106-byte region [0x60b00000a8b0,0x60b00000a91a)
allocated by thread T0 here:
#0 0x7f7e7a5f3602 in malloc (/usr/lib/x86_64-linux-gnu/libasan.so.2+0x98602)
#1 0x41e88d in splitline /home/ubuntu/mfc_fuzz/discount/markdown.c:174
#2 0x41e88d in htmlblock /home/ubuntu/mfc_fuzz/discount/markdown.c:333
SUMMARY: AddressSanitizer: heap-buffer-overflow ??:0 __asan_memmove
Shadow bytes around the buggy address:
0x0c167fff94d0: fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa
0x0c167fff94e0: fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa
0x0c167fff94f0: fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa
0x0c167fff9500: fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa
0x0c167fff9510: fa fa fa fa fa fa 00 00 00 00 00 00 00 00 00 00
=>0x0c167fff9520: 00 00 00[02]fa fa fa fa fa fa fa fa 00 00 00 00
0x0c167fff9530: 00 00 00 00 00 00 00 00 04 fa fa fa fa fa fa fa
0x0c167fff9540: fa fa fd fd fd fd fd fd fd fd fd fd fd fd fd fa
0x0c167fff9550: fa fa fa fa fa fa fa fa 00 00 00 00 00 00 00 00
0x0c167fff9560: 00 00 00 00 04 fa fa fa fa fa fa fa fa fa 00 00
0x0c167fff9570: 00 00 00 00 00 00 00 00 00 00 04 fa fa fa fa fa
Shadow byte legend (one shadow byte represents 8 application bytes):
Addressable: 00
Partially addressable: 01 02 03 04 05 06 07
Heap left redzone: fa
Heap right redzone: fb
Freed heap region: fd
Stack left redzone: f1
Stack mid redzone: f2
Stack right redzone: f3
Stack partial redzone: f4
Stack after return: f5
Stack use after scope: f8
Global redzone: f9
Global init order: f6
Poisoned by user: f7
Container overflow: fc
Array cookie: ac
Intra object redzone: bb
ASan internal: fe
==4515==ABORTING
The input test case is at: https://github.com/fCorleone/fuzz_programs/blob/master/discount/issue1_testcase
The executable file is mkd2html.
The text was updated successfully, but these errors were encountered: