[9fans] acid analogues of gdb stepping commands

[9fans] acid analogues of gdb stepping commands

[Nick Owens]

[-- Attachment #1: Type: text/plain, Size: 479 bytes --]

9fans,

i find myself needing to use acid a lot lately, and maybe i am going
about it wrong. i very much dislike that when i next() in acid, acid
will step through every subcall. what i would like is that it would
instead behave more like gdb's next. is this crazy or am i going about
using acid wrong?

maybe someone has acid analogues of gdb's next, finish, until as
described here.

https://sourceware.org/gdb/onlinedocs/gdb/Continuing-and-Stepping.html

nick


[-- Attachment #2: Type: application/pgp-signature, Size: 851 bytes --]

Re: [9fans] acid analogues of gdb stepping commands

[Skip Tavakkolian]

Isn't step what you're looking for?

> On Nov 7, 2013, at 6:46 PM, Nick Owens <mischief@9.offblast.org> wrote:
>
> 9fans,
>
> i find myself needing to use acid a lot lately, and maybe i am going
> about it wrong. i very much dislike that when i next() in acid, acid
> will step through every subcall. what i would like is that it would
> instead behave more like gdb's next. is this crazy or am i going about
> using acid wrong?
>
> maybe someone has acid analogues of gdb's next, finish, until as
> described here.
>
> https://sourceware.org/gdb/onlinedocs/gdb/Continuing-and-Stepping.html
>
> nick
>

Re: [9fans] acid analogues of gdb stepping commands

[Skip Tavakkolian]

[-- Attachment #1: Type: text/plain, Size: 1188 bytes --]

ok, read this again and i see that you want to avoid all the intermediate
prints from next(). i looked at /sys/lib/acid/port but can't see where the
print is happening; maybe it's deeper in builtins.

something like this, but with better housekeeping to work with existing
acid code, could work:

defn Next() {
printto("/env/nextStmt", pcfile(*PC),":",pcline(*PC)+1);
bpset(filepc(readfile("/env/nextStmt")));
cont();
}


On Thu, Nov 7, 2013 at 11:40 PM, Skip Tavakkolian <
skip.tavakkolian@gmail.com> wrote:

> Isn't step what you're looking for?
>
> > On Nov 7, 2013, at 6:46 PM, Nick Owens <mischief@9.offblast.org> wrote:
> >
> > 9fans,
> >
> > i find myself needing to use acid a lot lately, and maybe i am going
> > about it wrong. i very much dislike that when i next() in acid, acid
> > will step through every subcall. what i would like is that it would
> > instead behave more like gdb's next. is this crazy or am i going about
> > using acid wrong?
> >
> > maybe someone has acid analogues of gdb's next, finish, until as
> > described here.
> >
> > https://sourceware.org/gdb/onlinedocs/gdb/Continuing-and-Stepping.html
> >
> > nick
> >
>

[-- Attachment #2: Type: text/html, Size: 2079 bytes --]

Re: [9fans] acid analogues of gdb stepping commands

[Nick Owens]

[-- Attachment #1: Type: text/plain, Size: 703 bytes --]

On Fri, Nov 08, 2013 at 02:03:59AM -0800, Skip Tavakkolian wrote:
> something like this, but with better housekeeping to work with existing
> acid code, could work:
> 
> defn Next() {
> printto("/env/nextStmt", pcfile(*PC),":",pcline(*PC)+1);
> bpset(filepc(readfile("/env/nextStmt")));
> cont();
> }

this seems like it should work, but sometimes the computed PC for the
next line is way off, and cont() just runs the entire program. even more
confusing, doing pcline of the computed pc prints the next line
correctly, e.g.

x = filepc(pcfile(*PC)+":"+itoa(pcline(*PC)+1));
print(pcline(x));

will show the next line, but the new PC in x is totally wrong.

any clue why?

nick


[-- Attachment #2: Type: application/pgp-signature, Size: 851 bytes --]

Re: [9fans] acid analogues of gdb stepping commands

[erik quanstrom]

> this seems like it should work, but sometimes the computed PC for the
> next line is way off, and cont() just runs the entire program. even more
> confusing, doing pcline of the computed pc prints the next line
> correctly, e.g.
>
> x = filepc(pcfile(*PC)+":"+itoa(pcline(*PC)+1));
> print(pcline(x));
>
> will show the next line, but the new PC in x is totally wrong.
>
> any clue why?

have you tried compiling with -N (no peephole opt).

- erik

Re: [9fans] acid analogues of gdb stepping commands

[Nick Owens]

[-- Attachment #1: Type: text/plain, Size: 952 bytes --]

On Fri, Nov 08, 2013 at 06:51:54PM -0500, erik quanstrom wrote:
> have you tried compiling with -N (no peephole opt).
> 
> - erik
> 

i hadn't, but i just tried it like so, where Next is basically as Skip described:

cpu% 8c -FTVw -N ps.c
cpu% 8l -N -o 8.ps ps.8
cpu% cp 8.ps /386/bin/ps
cpu% acid /bin/ps
/bin/ps:386 plan 9 executable
/sys/lib/acid/port
/sys/lib/acid/386
acid: new()
1645200: system call  _main SUBL  $0x48,SP
1645200: breakpoint main+0x3  MOVL  $0x1,none+0x2c(SP)
acid: Next()
mischief          1    0:00   0:00      220K Await    bootrc
mischief          2   12:39   0:00        0K Wakeme   genrandom
<rest of output elided>
1645200: breakpoint main+0xb  CMPL  argv0(SB),$0x0
<stdin>:3: (error) msg: pid=1645200 startstop: process exited
acid: 

Bsrc shows this happens right after ARGBEGIN. maybe that confuses acid?

i should mention now that i am on 9front, but i hope that doesn't matter.

nick


[-- Attachment #2: Type: application/pgp-signature, Size: 851 bytes --]