Re: Request for implementation of KERN_PROC_SIGTRAMP sysctl

[Jason Thorpe <thorpej%me.com@localhost>]

> On Nov 21, 2021, at 8:28 AM, Jason Thorpe <thorpej%me.com@localhost> wrote:
> 
> If strip removes it, then you’re doomed anyway and trampoline assist via a function won’t help you, because you won’t be able to get to the trampoline or past it anyway.

Here’s a before/after of a “strip -s” of a binary on the DWARF unwind information:

 12 .eh_frame_hdr 000001a4  00000001200077f0  00000001200077f0  000077f0  2**2
                  CONTENTS, ALLOC, LOAD, READONLY, DATA
 13 .eh_frame     000007a4  0000000120007998  0000000120007998  00007998  2**3
                  CONTENTS, ALLOC, LOAD, READONLY, DATA
 
 
 12 .eh_frame_hdr 000001a4  00000001200077f0  00000001200077f0  000077f0  2**2
                  CONTENTS, ALLOC, LOAD, READONLY, DATA
 13 .eh_frame     000007a4  0000000120007998  0000000120007998  00007998  2**3
                  CONTENTS, ALLOC, LOAD, READONLY, DATA

I.e. strip does not effect the unwind information, because unwind information is not debugging information nor is it part of the symbol information; it is, in fact, required for correct operation of the program in the face of exceptions.  And my test program still works:

alpha-vm:thorpej 22$ ./test1  
^Cx 2
Backtrace 4 stack frames.
0x1200014e4 <_init+0x304> at ./test1
0x3fffdddbfd4 <__sigtramp_siginfo_2> at /usr/lib/libc.so.12
0x1200015e4 <_init+0x404> at ./test1
0x120001590 <_init+0x3b0> at ./test1

As you can see, because I stripped the symbols out of the program binary, the symbol names are wrong, but the unwind works and the program counter values are the same as an un-stripped copy of the program:

alpha-vm:thorpej 23$ ./sigbttest                                               
^Cx 2
Backtrace 4 stack frames.
0x1200014e4 <handler+0x54> at ./sigbttest
0x3fffdddbfd4 <__sigtramp_siginfo_2> at /usr/lib/libc.so.12
0x1200015e4 <the_loop+0x1c> at ./sigbttest
0x120001590 <main+0x48> at ./sigbttest

So, I think your worry about it is unwarranted.

>> On Nov 21, 2021, at 7:54 AM, John Marino (NetBSD) <netbsd%marino.st@localhost> wrote:
>> 
>> 
>> I'm not very familiar with CFA information.  I've been worried that strip(1) removes those symbols.  Is that worry meritless?

-- thorpej