karl-zylinski
commented
5 months ago A simpler repro without the Maybe is this:
package bug_test
TwelveByteStruct :: struct {
thing1: i32,
thing2: i32,
thing3: i32,
}
main :: proc() {
test :: proc(tbs: TwelveByteStruct) {
}
test({})
}I think this means that any proc that takes a 12 byte param will cause a stack variable overflow.
Also "tarık b." on discord wrote about this a few weeks ago, which might be another variant of the same bug. It's also a 16 byte write into a 12 byte region... But when calling fmt.println(), I guess maybe the varargs are collected as a single block of memory and then written into the called procs stack, similar to the struct above?: (message link where they wrote this: https://discord.com/channels/568138951836172421/585072813954564100/1193291921414705203)
package prog
import "vendor:glfw"
import "core:fmt"
main :: proc() {
major, minor, rev := glfw.GetVersion()
fmt.printf("OpenGL version: %d.%d.%d\n", major, minor, rev)
}==19165==ERROR: AddressSanitizer: stack-buffer-overflow on address 0x7f328d500020 at pc 0x000000506011 bp 0x7ffe9aae4230 sp 0x7ffe9aae4228
WRITE of size 16 at 0x7f328d500020 thread T0
#0 0x506010 in prog.main odin_package
#1 0x509044 in main (.../prog/prog+0x509044) (BuildId: 07bef459965573272c3f2a0ce6a5d421dd5406b3)
#2 0x7f328f629d8f in __libc_start_call_main csu/../sysdeps/nptl/libc_start_call_main.h:58:16
#3 0x7f328f629e3f in __libc_start_main csu/../csu/libc-start.c:392:3
#4 0x42a2f4 in _start (.../prog/prog+0x42a2f4) (BuildId: 07bef459965573272c3f2a0ce6a5d421dd5406b3)
Address 0x7f328d500020 is located in stack of thread T0 at offset 32 in frame
#0 0x505f3f in prog.main odin_package
This frame has 9 object(s):
[32, 44) '' <== Memory access at offset 32 partially overflows this variable
[64, 68) 'major'
[80, 84) 'minor'
[96, 100) 'rev'
/*...*/
HINT: this may be a false positive if your program uses some custom stack unwind mechanism, swapcontext or vfork
(longjmp and C++ exceptions *are* supported)
SUMMARY: AddressSanitizer: stack-buffer-overflow odin_package in prog.main
Shadow bytes around the buggy address:
/*...*/
=>0x7f328d500000: f1 f1 f1 f1[00]04 f3 f3 00 00 00 00 00 00 00 00
/*...*/
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
Freed heap region: fd
Stack left redzone: f1
Stack mid redzone: f2
Stack right redzone: f3
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
Left alloca redzone: ca
Right alloca redzone: cb
==19165==ABORTING
laytan
gingerBill
JesseRMeyer
The follow program will, when compiled with
odin run . -sanitize:addressreport a 16 byte write in a 12 byte region:At first I was trying to see if this was a false positive. But according to @laytan on Discord it seems like the IR is trying to store
2 x i64 into a { [2 x i32], i32 }. In this example I use a Maybe of [2]i32, which means2xi32plusi32for the tag, i.e. 12 bytes, see first comment of the issue for a non-Maybe repro.The IR is doing this both with and without ASAN address sanitizer enabled. Details of that discussion here: https://discord.com/channels/568138951836172421/585072813954564100/1202389269449023510
odin reportoutput: Odin: dev-2024-01:eb5b86557 OS: Windows 10 Professional (version: 22H2), build 19045.3930 CPU: Intel(R) Core(TM) i7-6950X CPU @ 3.00GHz RAM: 32674 MiBFailure Information (for bugs)
ASAN error output:
Steps to Reproduce
odin build . -sanitize:address