Re: armv7 lang/gcc12 "no bootstrap" build via system clang 15.0.7 based poudriere build ends up stuck in a small loop

On Mar 7, 2023, at 03:45, Dimitry Andric <dim@FreeBSD.org> wrote:

> On 7 Mar 2023, at 11:26, Mark Millard <marklmi@yahoo.com> wrote:
>> 
>> Below is a small example C source showing the clang 15+ armv7
>> problem that leads to the unbounded looping in later code in
>> the lang/gcc12+ builds: a data structure is mis-initialized,
>> breaking its invariant properties used by the later code
>> structure.
> ...
>> In the output below, note the blocks of 4 "next"
>> values that do not change. Each should match the
>> earlier hexadecimal value on the same line: point
>> back to same element of the array. 3 of 4 do not.
>> 
>> # cc -g -O2 partition.c
>> # ./a.out
>> 0: 0x40a84 : next?: 0x40a84
>> 1: 0x40a90 : next?: 0x40a84
>> 2: 0x40a9c : next?: 0x40a84
>> 3: 0x40aa8 : next?: 0x40a84
> 
> Very strange, I don't see this at all here. All the values are always
> correct, at every optimization level. I tried clang 13, clang 14, clang
> 15 from base and clang 15 from ports. However, my architecture is amd64,
> so maybe you are running into a armv7 specific issue?
> 

Yep: aarch64 and amd64 gcc builds all worked fine
via system-clang 15 and the test case also works
fine in my context on those. (All platforms
have FreeBSD built from copies of the same
source code.)

It is only on armv7 that I got the gcc build
problems and the test case failures.

FYI:

aarch64 Fedora 37 using clang-15 also works
fine for me (only tried the test case, not
gcc builds):

# clang-15      -g -O2 partition.c ; ./a.out
0: 0x420038 : next?: 0x420038
1: 0x420048 : next?: 0x420048
2: 0x420058 : next?: 0x420058
3: 0x420068 : next?: 0x420068
4: 0x420078 : next?: 0x420078
5: 0x420088 : next?: 0x420088
6: 0x420098 : next?: 0x420098
7: 0x4200a8 : next?: 0x4200a8
8: 0x4200b8 : next?: 0x4200b8
9: 0x4200c8 : next?: 0x4200c8
10: 0x4200d8 : next?: 0x4200d8
11: 0x4200e8 : next?: 0x4200e8
12: 0x4200f8 : next?: 0x4200f8
13: 0x420108 : next?: 0x420108
14: 0x420118 : next?: 0x420118
15: 0x420128 : next?: 0x420128
16: 0x420138 : next?: 0x420138
17: 0x420148 : next?: 0x420148
18: 0x420158 : next?: 0x420158
19: 0x420168 : next?: 0x420168
20: 0x420178 : next?: 0x420178
21: 0x420188 : next?: 0x420188
22: 0x420198 : next?: 0x420198
23: 0x4201a8 : next?: 0x4201a8
24: 0x4201b8 : next?: 0x4201b8
25: 0x4201c8 : next?: 0x4201c8
26: 0x4201d8 : next?: 0x4201d8
27: 0x4201e8 : next?: 0x4201e8
28: 0x4201f8 : next?: 0x4201f8
29: 0x420208 : next?: 0x420208
30: 0x420218 : next?: 0x420218
31: 0x420228 : next?: 0x420228

# clang-15 -v
clang version 15.0.7 (Fedora 15.0.7-1.fc37)
Target: aarch64-redhat-linux-gnu
Thread model: posix
InstalledDir: /usr/bin
Found candidate GCC installation: /usr/bin/../lib/gcc/aarch64-redhat-linux/12
Selected GCC installation: /usr/bin/../lib/gcc/aarch64-redhat-linux/12
Candidate multilib: .;@m64
Selected multilib: .;@m64

If I understand right, Lorenzo indicates that he has
more failing contexts. I've no evidence of my own of
such.

===
Mark Millard
marklmi at yahoo.com