[PATCH] D28386: Add the ability to iterate across all attributes in a DIE.

[Greg Clayton via llvm-commits]

> On Jan 9, 2017, at 2:41 PM, David Blaikie <dblaikie at gmail.com> wrote:
> 
> 
> 
> On Mon, Jan 9, 2017 at 2:29 PM Greg Clayton <clayborg at gmail.com <mailto:clayborg at gmail.com>> wrote:
>> On Jan 9, 2017, at 2:23 PM, David Blaikie <dblaikie at gmail.com <mailto:dblaikie at gmail.com>> wrote:
>> 
>> I'm not sure optional error handling's ideal - seems errors should always be handled (which is the strong premise behind Lang's work introducing llvm::Error).
> 
> What is a DWARF parser during a debug session gonna do with any of these errors? Emit them in the console? If we have borked DWARF we can’t just emit every error we come across, the user will not like that.
> 
> Quite possibly, yes. That seems to be what LLDB does already for DWARF it can't handle - and I imagine not reading any attributes would start to look like "not handling" to a user when things show up without names, address ranges, or any other functionality?

I guess after we parse the DWARF and make the skeleton of the DIEs successfully that the input is good and that is kind of an invariant. We wouldn’t be extracting any data from DIEs using a DWARFDie if we weren’t able to parse the debug info already to get the layout of the DWARF. I would like to assume that the DWARF is good at this point and we are just accessing it.

>  
> If we add error handling, I don’t really plan on integrating it into the LLDB parser, but you seemed to want this so I added it. I personally don’t think it is useful.
> 
> What should LLDB do when it encounters broken input? What does it do when it encounters input it can't parse today?

Again, the DWARF would have failed to parse and no DIEs would be visited if the pass through the compile units died in the first place. We just try to iterate through the attributes and if anything goes wrong we exit the attribute loop. I really don’t see the point of the errors, or if they have to exist then they should be optional.

Greg

>  
> 
> Greg
> 
>> On Mon, Jan 9, 2017 at 2:21 PM Greg Clayton via Phabricator <reviews at reviews.llvm.org <mailto:reviews at reviews.llvm.org>> wrote:
>> clayborg updated this revision to Diff 83698.
>> clayborg added a comment.
>> 
>> Added optional error handling to the attributes iterators. You can now pass an "llvm::Error *" to the attributes():
>> 
>>   attribute_iterator DWARFDie::attributes(llvm::Error *Err);
>> 
>> If the error is non-NULL, then the error will be filled in. This iteration error is the same method used in Archive.h/Archive.cpp after I spoke with Lang Hames.
>> 
>> 
>> https://reviews.llvm.org/D28386 <https://reviews.llvm.org/D28386>
>> 
>> Files:
>>   include/llvm/DebugInfo/DWARF/DWARFAbbreviationDeclaration.h
>>   include/llvm/DebugInfo/DWARF/DWARFAttribute.h
>>   include/llvm/DebugInfo/DWARF/DWARFDie.h
>>   lib/DebugInfo/DWARF/DWARFDie.cpp
>>   unittests/DebugInfo/DWARF/DWARFDebugInfoTest.cpp
>> 
> 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20170109/91778e31/attachment.html>