Hi Alex and all,
We should rationalize our calls to subroutine grid2deg() to make them
acceptable to the increasingly fussy versions of gfortran. This will
not be done in time for WSJT-X 2.5.4, which will be released very soon,
but probably in the next release after that.
-- 73, Joe, K1JT
On 12/27/2021 7:16 PM, Alex Lelievre via wsjt-devel wrote:
Unfortunately I just discovered a bug that crashes WSJT-X running native
on Apple M1 when you double-click a call in the Band Activity window.
This is the error in case anyone has seen something similar:
At line 1 of file
/Users/alex/wsjtx-debug/wsjtx-prefix/src/wsjtx/lib/grid2deg.f90
Fortran runtime error: Actual string length is shorter than the declared
one for dummy argument 'grid0' (3/6)
libc++abi: terminating with uncaught exception of type
boost::wrapexcept<boost::exception_detail::error_info_injector<boost::log::v2_mt_posix::system_error>
>: Failed to set TLS value: Invalid argument
Process 94887 stopped
* thread #10, name = 'QThread', stop reason = signal SIGABRT
frame #0: 0x00000001b10499b8libsystem_kernel.dylib`__pthread_kill + 8
libsystem_kernel.dylib`__pthread_kill:
-> 0x1b10499b8 <+8>: b.lo 0x1b10499d8 ; <+40>
0x1b10499bc <+12>: pacibsp
0x1b10499c0 <+16>: stp x29, x30, [sp, #-0x10]!
0x1b10499c4 <+20>: mov x29, sp
Target 0: (wsjtx) stopped.
alex
K6LOT
On Dec 27, 2021, at 1:11 PM, Alex Lelievre <a...@foinc.com
<mailto:a...@foinc.com>> wrote:
I wanted to post a progress update to getting WSJT-X built and running
natively on Apple M1 hardware.
Today I managed to get FT8 to work natively on Apple M1 with version
2.5.3. Previously this was crashing due to a compiler issue (stacks
on M1 are not executable). The gcc community posted a fix for this
issue a few days ago and amazingly now the Jt9 process runs without
crashing!! Amazing!
There are still some remaining issues that I had to fix up manually
for those that are interested:
- I had to stub __chkstk_darwin()
- I had to replace libgcc_s.1.1.dylib with libgcc_s.dylib (and then
renamed to libgcc_s.1.1.dylib to cheat)… This is necessary otherwise
___addtf3 is an unresolved import and causes WSJT-X to crash on launch.
- I had to manually sign the executable to run on Monterey 12.1.
If anyone else is interested in building a native version of WSJT-X
for Apple M1 please let me know and I will post my instructions… Once
those issues listed above are resolved, I can look into getting this
build into the official releases (which I have no idea how to do).
Also just wanted to express my deepest condolences to Bill
Somerville’s family and friends. If it wasn’t for his help I would
never have been able to get the code ported over to Apple M1. He
helped me learn Fortran and answered my deluge of questions with
enthusiasm and humor. I will truly miss his support. RIP
alex
K6LOT
On Aug 27, 2021, at 10:25 AM, Alex Lelievre <a...@foinc.com
<mailto:a...@foinc.com>> wrote:
Good thing I didn’t see that Qt blog article otherwise I wouldn’t
have tried to build for arm64. ;-)
I was holding out until I saw that most of brew now supported the M1.
I am still having some issues with the jt9 submodule crashing so I
have some more work to do getting the fortran code to play nice with
gfortran 11.x.
I suspect I can fix this crash if I quiet some warnings about
exceeding stack sizes and/or reentrancy warnings…
<Screen Shot 2021-08-27 at 8.46.41 AM.png>
Alex
[snip]
_______________________________________________
wsjt-devel mailing list
wsjt-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/wsjt-devel
_______________________________________________
wsjt-devel mailing list
wsjt-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/wsjt-devel
