Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Failing test - strings.lua #39

Open
Pavel-Durov opened this issue Aug 14, 2023 · 2 comments
Open

Failing test - strings.lua #39

Pavel-Durov opened this issue Aug 14, 2023 · 2 comments

Comments

@Pavel-Durov
Copy link
Contributor

Issue

Error on running staings.lua test with both serial and non-serial compilation:

$ YKD_SERIALISE_COMPILATION=0 try_repeat 1000 ../src/lua -e"_U=true" strings.lua
.....
testing strings and string library
testing 'format %a %A'
'collate' locale not found
'ctype' locale not found

 >>> testC not active: skipping 'pushfstring' tests <<<

OK
Segmentation fault (core dumped)
$ YKD_SERIALISE_COMPILATION=1 try_repeat 1000 ../src/lua -e"_U=true" strings.lua
testing strings and string library
testing 'format %a %A'
'collate' locale not found
'ctype' locale not found

 >>> testC not active: skipping 'pushfstring' tests <<<

OK
realloc(): invalid pointer
Aborted (core dumped)

Notes

Both compilation modes are non-deterministic - sometimes it errors out and sometimes completes successfully.

Versions:

YkLua - yk-location-allocation/056cf6ae7cb3ebd59a7e0dc89df99c03f0e8d3d9
Yk - master/a690e6eaf7253a4fa9c04bb54406c07b7efc0085

bors bot added a commit that referenced this issue Aug 25, 2023
64: Track initialised yk locations r=ltratt a=Pavel-Durov

Partially fixes (only for serial compilation)  - #62

# Changes

- Remove reallocarray in favour of calloc
- Change YkLocations to be stored as pointers
- Change lyk interface to reflect its "hooks" functionality
- Move tests to a separate test script and enable serialised all.lua test suite
- Added logging in lyk module for ease of debugging
- Updated readme

`test.sh` was tested for resiliency multiple times:
```shell
$ try_repeat -v 100 sh ./test.sh
``` 
# Issues

## Uninitialised locations memory 
`reallocarray` does not initialise memory with default zero bytes locations as `calloc`. 
Moving to `calloc` and using pointers allowed to check for NULL with default values set at initialisation time.

## Tests

When lua tests are executed one by one as single files it produces different results from when it's executed via `all.lua` test suite.

Example:

```shell

$ YKD_SERIALISE_COMPILATION=1 ../src/lua -e"_U=true" ./main.lua 
...
../src/lua: ./main.lua:343: assertion failed!
stack traceback:
        [C]: in function 'assert'
        ./main.lua:343: in main chunk
        [C]: in ?
```
When run through `all.lua` it passes:
```shell
$ YKD_SERIALISE_COMPILATION=1 ../src/lua -e"_U=true" ./all.lua 
...
***** FILE 'main.lua'*****
...
***** FILE 'gc.lua'*****
...
```
Since `all.lua` is what we base the stability of yklua, I think we should include it in the tests as is, currently only serialised compilation works.

Running `all.lua` prior to these changes results in error (`main/56c5787799b876f36babbae24e9afc025806b831`):
```
$ YKD_SERIALISE_COMPILATION=1 gdb -ex 'r' -ex 'bt' --args ../src/lua -e"_U=true" ./all.lua 
...
***** FILE 'main.lua'*****

Program received signal SIGSEGV, Segmentation fault.
core::sync::atomic::AtomicUsize::fetch_sub (self=<optimised out>) at /rustc/ef85656a10657ba5e4f7fe2931a4ca6293138d51/library/core/src/sync/atomic.rs:2575
2575    /rustc/ef85656a10657ba5e4f7fe2931a4ca6293138d51/library/core/src/sync/atomic.rs: No such file or directory.
#0  core::sync::atomic::AtomicUsize::fetch_sub (self=<optimised out>) at /rustc/ef85656a10657ba5e4f7fe2931a4ca6293138d51/library/core/src/sync/atomic.rs:2575
#1  alloc::sync::{impl#33}::drop<lock_api::mutex::Mutex<parking_lot::raw_mutex::RawMutex, ykrt::location::HotLocation>, alloc::alloc::Global> (self=0x7fffffffb410)
    at /rustc/ef85656a10657ba5e4f7fe2931a4ca6293138d51/library/alloc/src/sync.rs:2370
#2  0x00007ffff7b0d4ab in core::ptr::drop_in_place<alloc::sync::Arc<lock_api::mutex::Mutex<parking_lot::raw_mutex::RawMutex, ykrt::location::HotLocation>, alloc::alloc::Global>> () at /rustc/ef85656a10657ba5e4f7fe2931a4ca6293138d51/library/core/src/ptr/mod.rs:497
#3  0x00007ffff7b0067e in core::mem::drop<alloc::sync::Arc<lock_api::mutex::Mutex<parking_lot::raw_mutex::RawMutex, ykrt::location::HotLocation>, alloc::alloc::Global>> (_x=...) at /rustc/ef85656a10657ba5e4f7fe2931a4ca6293138d51/library/core/src/mem/mod.rs:987
#4  0x00007ffff7b121f0 in ykrt::location::{impl#1}::drop (self=0x7fffffffb468) at ykrt/src/location.rs:197
#5  0x00007ffff7affa9b in core::ptr::drop_in_place<ykrt::location::Location> () at /rustc/ef85656a10657ba5e4f7fe2931a4ca6293138d51/library/core/src/ptr/mod.rs:497
#6  0x00007ffff7aff6bd in core::mem::drop<ykrt::location::Location> (_x=...) at /rustc/ef85656a10657ba5e4f7fe2931a4ca6293138d51/library/core/src/mem/mod.rs:987
#7  0x00007ffff7b004bd in ykcapi::yk_location_drop (loc=...) at ykcapi/src/lib.rs:90
#8  0x000000000088aa0e in free_loc (f=<optimised out>, i=<optimised out>, idx=<optimised out>) at lyk.c:66
#9  0x000000000088aba5 in yk_free_locactions (f=0x928cc0) at lyk.c:76
--Type <RET> for more, q to quit, c to continue without paging--
#10 0x0000000000807738 in luaF_freeproto (L=0x916e38, f=0x928cc0) at lfunc.c:276
#11 0x0000000000809fde in freeobj (L=0x916e38, o=0x928cc0) at lgc.c:767
#12 0x0000000000817145 in sweepgen (L=0x916e38, g=<optimised out>, p=<optimised out>, limit=<optimised out>, pfirstold1=<optimised out>) at lgc.c:1106
#13 0x0000000000816713 in youngcollection (L=0x916e38, g=0x916f00) at lgc.c:1239
#14 0x000000000081557d in genstep (L=0x916e38, g=0x916f00) at lgc.c:1434
#15 0x0000000000815044 in luaC_step (L=0x916e38) at lgc.c:1686
#16 0x00000000007e4447 in lua_pushstring (L=0x916e38, s=<optimised out>) at lapi.c:553
#17 0x000000000089e60c in findloader (L=0x916e38, name=0x928df8 "tracegc") at loadlib.c:641
#18 0x000000000089de0a in ll_require (L=0x916e38) at loadlib.c:666
#19 0x00000000007fda75 in precallC (L=0x916e38, func=<optimised out>, nresults=<optimised out>, f=0x89db80 <ll_require>) at ldo.c:506
#20 0x00000000007fe016 in luaD_precall (L=0x916e38, func=0x923780, nresults=1) at ldo.c:569
#21 0x0000000000883f68 in luaV_execute (L=0x916e38, ci=<optimised out>) at lvm.c:1655
#22 0x00000000007feb3b in ccall (L=0x916e38, func=<optimised out>, nResults=<optimised out>, inc=<optimised out>) at ldo.c:609
#23 0x00000000007fec61 in luaD_callnoyield (L=0x916e38, func=0x917730, nResults=-1) at ldo.c:627
#24 0x00000000007eaf03 in f_call (L=0x916e38, ud=<optimised out>) at lapi.c:1041
#25 0x00000000007f8be7 in luaD_rawrunprotected (L=0x916e38, f=0x7eae40 <f_call>, ud=0x7ffff2487308) at ldo.c:144
#26 0x0000000000801406 in luaD_pcall (L=0x916e38, func=0x7eae40 <f_call>, u=0x7ffff2487308, old_top=<optimised out>, ef=<optimised out>) at ldo.c:926
#27 0x00000000007ea9ec in lua_pcallk (L=0x916e38, nargs=<optimised out>, nresults=<optimised out>, errfunc=<optimised out>, ctx=<optimised out>, k=<optimised out>)
    at lapi.c:1067
#28 0x00000000007dc623 in docall (L=0x916e38, narg=0, nres=-1) at lua.c:160
#29 0x00000000007dbd24 in handle_script (L=0x916e38, argv=<optimised out>) at lua.c:255
#30 0x00000000007d9fe3 in pmain (L=0x916e38) at lua.c:634
#31 0x00000000007fda75 in precallC (L=0x916e38, func=<optimised out>, nresults=<optimised out>, f=0x7d97f0 <pmain>) at ldo.c:506
#32 0x00000000007fe0c8 in luaD_precall (L=0x916e38, func=0x9176f0, nresults=1) at ldo.c:572
#33 0x00000000007fea7f in ccall (L=0x916e38, func=0x9176f0, nResults=1, inc=<optimised out>) at ldo.c:607
#34 0x00000000007fec61 in luaD_callnoyield (L=0x916e38, func=0x9176f0, nResults=1) at ldo.c:627
#35 0x00000000007eaf03 in f_call (L=0x916e38, ud=<optimised out>) at lapi.c:1041
#36 0x00000000007f8be7 in luaD_rawrunprotected (L=0x916e38, f=0x7eae40 <f_call>, ud=0x7ffff2487058) at ldo.c:144
#37 0x0000000000801406 in luaD_pcall (L=0x916e38, func=0x7eae40 <f_call>, u=0x7ffff2487058, old_top=<optimised out>, ef=<optimised out>) at ldo.c:926
#38 0x00000000007ea9ec in lua_pcallk (L=0x916e38, nargs=<optimised out>, nresults=<optimised out>, errfunc=<optimised out>, ctx=<optimised out>, k=<optimised out>)
    at lapi.c:1067
#39 0x00000000007d94b0 in main (argc=<optimised out>, argv=<optimised out>) at lua.c:660
```

That's cause we're freeing uninitialised yk locations.


Co-authored-by: Pavel Durov <[email protected]>
@Pavel-Durov
Copy link
Contributor Author

Pavel-Durov commented Oct 18, 2023

Error update:

YKLUA - 950a7ab
YK - c6a0b9ca333408693a0c904bd502b1cdf0dc8d3f

$ YKD_SERIALISE_COMPILATION=1 gdb -batch -ex r -ex bt --args  ../src/lua strings.lua

 >>> testC not active: skipping 'pushfstring' tests <<<

OK

Thread 6 "lua" received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0x7ffff1afe6c0 (LWP 1328069)]
__memcpy_avx_unaligned_erms () at ../sysdeps/x86_64/multiarch/memmove-vec-unaligned-erms.S:664
664     ../sysdeps/x86_64/multiarch/memmove-vec-unaligned-erms.S: No such file or directory.
#0  __memcpy_avx_unaligned_erms () at ../sysdeps/x86_64/multiarch/memmove-vec-unaligned-erms.S:664
#1  0x00007ffff7b122e5 in read_aux (aux_buf=0x7fffe8000000, hdr=0x7ffff243f000, trace=0x38f1900, err=0x36ab938) at src/perf/collect.c:296
#2  0x00007ffff7b12165 in handle_sample (aux_buf=0x7fffe8000000, hdr=0x7ffff243f000, trace=0x38f1900, data_tmp=0x7fffed033440, err=0x36ab938) at src/perf/collect.c:241
#3  0x00007ffff7b11f0c in poll_loop (perf_fd=5, stop_fd=3, mmap_hdr=0x7ffff243f000, aux=0x7fffe8000000, trace=0x38f1900, err=0x36ab938) at src/perf/collect.c:353
#4  0x00007ffff7b11bf0 in collector_thread (arg=0x7fffffffb9e8) at src/perf/collect.c:480
#5  0x00007ffff78a8044 in start_thread (arg=<optimized out>) at ./nptl/pthread_create.c:442
#6  0x00007ffff792861c in clone3 () at ../sysdeps/unix/sysv/linux/x86_64/clone3.S:81

The YKD_SERIALISE_COMPILATION=0 runtime error is non-deterministic, also I couldn't reproduce it with gdb attached therefore I can't provide a backtrace.

@Pavel-Durov
Copy link
Contributor Author

Update

ERROR - YKD_SERIALISE_COMPILATION=1
ERROR - YKD_SERIALISE_COMPILATION=0

Versions

YKLUA - main/810108764e996d727442d6984f7a8318155a08c8
YK - master/a6613631e8e13cfe3ba4cc8ee24df6517831fd6c

Backtrace

YKD_SERIALISE_COMPILATION=0 try_repeat -v 1000  ../src/lua -e"_U=true" ./strings.lua
...
testing strings and string library
testing 'format %a %A'
'collate' locale not found
'ctype' locale not found

 >>> testC not active: skipping 'pushfstring' tests <<<

OK
double free or corruption (!prev)
Aborted
YKD_SERIALISE_COMPILATION=1 try_repeat -v 1000  ../src/lua -e"_U=true" ./strings.lua
===> 1: ../src/lua -e_U=true ./strings.lua
testing strings and string library
testing 'format %a %A'
'collate' locale not found
'ctype' locale not found

 >>> testC not active: skipping 'pushfstring' tests <<<

OK
Segmentation fault

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

No branches or pull requests

1 participant