Bug 2354997

gdb 16's gstack now unconditionally passes --readnever to gdb which omits most of the useful information that I've come to rely, in particular the file and line number.

There doesn't appear to be a way to override this. gstack supports $GDBARGS but that cannot be used to override the hardcoded --readnever.

Test program to illustrate the case is attached - it forks off a worker process and then gstack to print the stacktrace of that process. This resembles the real-world use-case I've had in the libinput test suite for the last 10y or so. The output of that program on three different versions is:

Fedora 41 with gdb-15.1-1.fc41.x86_64:

#0  0x00007f79c7bc1117 in clock_nanosleep.5 () from /lib64/libc.so.6
#1  0x00007f79c7bcd247 in nanosleep () from /lib64/libc.so.6
#2  0x00007f79c7bdf8b1 in sleep () from /lib64/libc.so.6
#3  0x00000000004011a1 in tired () at test.c:9
#4  0x00000000004011ca in main () at test.c:16

Fedora 41 with gdb-16.2-1.fc41.x86_64:

Thread 1 (Thread 0x7fae7acff740 (LWP 52285) "test"):
#0  0x00007fae7adbf117 in clock_nanosleep.5 () from /lib64/libc.so.6
#1  0x00007fae7adcb247 in nanosleep () from /lib64/libc.so.6
#2  0x00007fae7addd8b1 in sleep () from /lib64/libc.so.6
#3  0x0000000038bd3536 in ?? ()
#4  0x0000000001000000 in ?? ()
#5  0x2a736384ed9f5000 in ?? ()
#6  0x00007ffd77630598 in ?? ()
#7  0x0000000000000001 in ?? ()
#8  0x00007ffd77630440 in ?? ()
#9  0x00000000004011a1 in tired ()

gdb 16 cannot resolve the function name but nothing else, gdb 15 still has gstack without --readnever so that provides all the information I need - files and line numbers.

Reproducible: Always

Steps to Reproduce:
1. gcc -ggdb -Wall -o test test.c
2. sudo ./test
Actual Results:  
gdb 16 omits any file names and line numbers

Expected Results:  
file names/line numbers included, or at least included upon passing some option

If --readnever is a requirement for some use cases it should IMO be toggle-able because right now gstack lost a good part of its rather convenient functionality.