From: Daniel Jacobowitz <drow@false.org>
To: Michael Snyder <msnyder@redhat.com>
Cc: gdb-patches@sources.redhat.com
Subject: Re: [rfa] Dwarf2 testsuite
Date: Fri, 08 Oct 2004 22:16:00 -0000 [thread overview]
Message-ID: <20041008221614.GA1584@nevyn.them.org> (raw)
In-Reply-To: <4166FAD8.9090301@redhat.com>
On Fri, Oct 08, 2004 at 01:38:48PM -0700, Michael Snyder wrote:
> Daniel Jacobowitz wrote:
> >This patch adds a sample test for DWARF-2 unit testing. It requires:
> > - gas, for the .uleb128/.sleb128 directives among other things
> > - ELF; it could probably be made to work on non-ELF dwarf2 systems
> > but I don't know what changes would be necessary. Perhaps those
> > should be tested separately.
> >
> >I haven't tried, but I wrote it with the intent that it would not care
> >about
> >the host beyond that; it should run on 64-bit ELF targets OK.
> >
> >It just includes a trivial test containing hand-written DIEs for a single
> >CU. This one file I annotated exhaustively, and let me tell you, it was
> >exhausting; for future tests I will allow them to be self-documenting (the
> >comments don't say anything that compiling the file and running readelf
> >won't tell you).
> >
> >The test itself just demonstrates that the dwarf2 reader isn't completely
> >busted. But more interesting tests could be easily added.
> >
> >Look OK? Suggestions?
> >
>
> Looks exhausting. I'm glad you're doing this.
>
> Dan, what would you think about (somebody) writing a tool
> that could read some sort of description (eg. what's in
> your comments), and generate the dwarf2 recs? Sort of a
> simple compiler that accepts, in effect, the dwarf2 spec
> as input? Would that be, perhaps, less of an effort than
> writing a sufficiently large subset of tests by hand?
<chanting> Do it! Do it! Do it! </chanting>
Does that answer your question? Yes, it would absolutely be less
effort. It would be a little tricky. Done properly could
encapsulate most of the time-consuming bits - in particular, it could
handle testing both 32-bit and 64-bit mode, and it could automate (for
the common cases at least) the nasty bits of .debug_abbrev and
.debug_line.
I'm not sure what it would output. Maybe a text file full of .byte
directives, so that we could #include it from an asm file with the
labels referenced in it.
--
Daniel Jacobowitz
prev parent reply other threads:[~2004-10-08 22:16 UTC|newest]
Thread overview: 9+ messages / expand[flat|nested] mbox.gz Atom feed top
2004-10-08 18:35 Daniel Jacobowitz
2004-10-08 19:45 ` Michael Chastain
2004-10-08 20:25 ` Daniel Jacobowitz
2004-10-11 3:50 ` Michael Chastain
2004-10-12 14:42 ` Daniel Jacobowitz
2004-10-12 14:44 ` Daniel Jacobowitz
2004-10-12 15:54 ` Michael Chastain
2004-10-08 20:38 ` Michael Snyder
2004-10-08 22:16 ` Daniel Jacobowitz [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20041008221614.GA1584@nevyn.them.org \
--to=drow@false.org \
--cc=gdb-patches@sources.redhat.com \
--cc=msnyder@redhat.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox