I never expected it was the number files, I give the number to show that it was
some what of general error that populated lost+found
Using lldb
(lldb) run -l
run -l
Process 89184 launched: '/bin/ls' (x86_64)
total 2176
c--------x 1 34078976 wheel 58, 7079544 Dec 31 1969 #04963796
Process 89184 stopped
* thread #1, stop reason = signal SIGSEGV
frame #0: 0x00000ae35b5aef18 ls`___lldb_unnamed_symbol508 + 1432
ls`___lldb_unnamed_symbol508:
-> 0xae35b5aef18 <+1432>: movl 0x10(%r11), %ecx
0xae35b5aef1c <+1436>: leaq -0x42c8b(%rip), %rax
0xae35b5aef23 <+1443>: cmpq $0xb, %rcx
0xae35b5aef27 <+1447>: ja 0xae35b5aef34 ; <+1460>
Process 89184 stopped
* thread #1: tid = 185603, 0x00000ae35b5aef18 ls`___lldb_unnamed_symbol508 +
1432, stop reason = signal SIGSEGV
(lldb) thread backtrace
thread backtrace
* thread #1, stop reason = signal SIGSEGV
* frame #0: 0x00000ae35b5aef18 ls`___lldb_unnamed_symbol508 + 1432
frame #1: 0x00000ae35b5ae93a ls`___lldb_unnamed_symbol507 + 90
frame #2: 0x00000ae35b578111 ls`___lldb_unnamed_symbol26 + 993
frame #3: 0x00000ae35b577ac9 ls`___lldb_unnamed_symbol22 + 1353
frame #4: 0x00000ae35b57745f ls`___lldb_unnamed_symbol20 + 495
frame #5: 0x00000ae35b57721f ls`___lldb_unnamed_symbol19 + 1887
frame #6: 0x00000ae35b5764c2 ls`___lldb_unnamed_symbol5 + 338
(lldb)
-----Original Message-----
From: Omar Polo <o...@omarpolo.com>
Sent: Tuesday, March 26, 2024 1:37 PM
To: Peter Fraser <p...@thinkage.ca>
Cc: bugs@openbsd.org
Subject: Re: ls -l Segmentation fault
[moving to bugs@]
On 2024/03/26 17:15:28 +0000, Peter Fraser <p...@thinkage.ca> wrote:
> For some reason I don't know my partition /var/www got screwed up and there
> was a long list in /var/www/lost+found.
>
> Since I populate /var/www by using rsync from another computer, and I expect
> didn't notice the existence of lost+found for a long time.
>
> "ls" on the directory works but "ls -l" faults with
>
> ls -l
> total 2176
> c--------x 1 34078976 wheel 58, 7079544 Dec 31 1969 #04963796
> Segmentation fault (core dumped)
>
> The directory has 1324 files in it.
>
> I have included the core dump of "ls"
>
> If you try to tar the files in the directory, you get very strange message,
> but the files are ignored and the is not faults.
> I still have them if anyone wants to examine them. If not, I will try to
> delete them.
the corefile alone is not going to be of much help due to the random libc
relinking; can you show the backtrace? lldb in base should work, otherwise use
egdb from the gdb package.
$ ls -l
[assuming it crashes]
$ egdb ls ls.core
[useless copyright stuff elided]
(gdb) bt
and paste the output in a mail. re-compiling ls with debug symbols (cd
/usr/src/bin/ls && make DEBUG='-g' && doas make DEBUG='-g' install) could be
useful too.
For what is worth, I have a maildir with 14k files and haven't noticed crashes,
so the number of entries it's probably not the culprit.
Thanks,
