[prev in list] [next in list] [prev in thread] [next in thread] 

List:       haiku-commits
Subject:    [haiku-commits] Re: haiku: hrev57242 - src/kits/debugger/dwarf
From:       Rene Gollent <anevilyak () gmail ! com>
Date:       2023-08-26 1:33:22
Message-ID: CAHVFsOh8YrXSe-NqQmx4Y4SYUYx0LJT6n2FwCoRFM2aT+01VJg () mail ! gmail ! com
[Download RAW message or body]

From a look at the spec, the only noteworthy difference I see in the line
number section is some added data to assisted with profile guided
optimization. We can probably ignore that and be fine.

Regards,

Rene

On Fri, Aug 25, 2023, 17:31 waddlesplash <waddlesplash@gmail.com> wrote:

> On Fri, Aug 25, 2023 at 5:18 PM Rene Gollent <anevilyak@gmail.com> wrote:
> >> +       if (version != 2 && version != 3) {
> >> +               WARNING("DwarfFile::_ParseLineInfo(\"%s\"): unsupported
> "
> >> +                       "version %d\n", fName, version);
> >> +               return B_UNSUPPORTED;
> >> +       }
> >> +
> >>         // header_length (4/8)
> >
> >
> > I'm mildly puzzled by this ; I could have sworn I updated us to handle
> dwarf version 4. I don't have the spec on hand, but does the version in the
> line number program section not correlate?
>
> The rest of the file handles DWARF4, but this one function didn't (you
> can see I just moved this version check up from lower down.) For
> whatever reason, GCC up through 11 generated version 2 or 3 line-info
> with DWARF4, but now it actually generates version 4 lineinfo for
> DWARF4.
>
> I took a look at LLDB's code related to this, and it looks like v4
> doesn't differ much (if any?) while v5 line-info differs
> significantly. I guess I could test adding version 4 in here and
> seeing what happens...
>
> -waddlesplash
>
>

[Attachment #3 (text/html)]

<div dir="auto">From a look at the spec, the only noteworthy difference I see in the \
line number section is some added data to assisted with profile guided optimization. \
We can probably ignore that and be fine.<div dir="auto"><br></div><div \
dir="auto">Regards,</div><div dir="auto"><br></div><div \
dir="auto">Rene</div></div><br><div class="gmail_quote"><div dir="ltr" \
class="gmail_attr">On Fri, Aug 25, 2023, 17:31 waddlesplash &lt;<a \
href="mailto:waddlesplash@gmail.com" target="_blank" \
rel="noreferrer">waddlesplash@gmail.com</a>&gt; wrote:<br></div><blockquote \
class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc \
solid;padding-left:1ex">On Fri, Aug 25, 2023 at 5:18 PM Rene Gollent &lt;<a \
href="mailto:anevilyak@gmail.com" rel="noreferrer noreferrer" \
target="_blank">anevilyak@gmail.com</a>&gt; wrote:<br> &gt;&gt; +           if \
(version != 2 &amp;&amp; version != 3) {<br> &gt;&gt; +                       \
WARNING(&quot;DwarfFile::_ParseLineInfo(\&quot;%s\&quot;): unsupported &quot;<br> \
&gt;&gt; +                                   &quot;version %d\n&quot;, fName, \
version);<br> &gt;&gt; +                       return B_UNSUPPORTED;<br>
&gt;&gt; +           }<br>
&gt;&gt; +<br>
&gt;&gt;              // header_length (4/8)<br>
&gt;<br>
&gt;<br>
&gt; I&#39;m mildly puzzled by this ; I could have sworn I updated us to handle dwarf \
version 4. I don&#39;t have the spec on hand, but does the version in the line number \
program section not correlate?<br> <br>
The rest of the file handles DWARF4, but this one function didn&#39;t (you<br>
can see I just moved this version check up from lower down.) For<br>
whatever reason, GCC up through 11 generated version 2 or 3 line-info<br>
with DWARF4, but now it actually generates version 4 lineinfo for<br>
DWARF4.<br>
<br>
I took a look at LLDB&#39;s code related to this, and it looks like v4<br>
doesn&#39;t differ much (if any?) while v5 line-info differs<br>
significantly. I guess I could test adding version 4 in here and<br>
seeing what happens...<br>
<br>
-waddlesplash<br>
<br>
</blockquote></div>



[prev in list] [next in list] [prev in thread] [next in thread]