[llvm-commits] [patch] Identifying code sections in MCJIT

Kaylor, Andrew andrew.kaylor at intel.com
Thu Dec 13 14:30:44 PST 2012


Hi Tim,

I just asked about this on IRC and I think I now know how to make isSectionText work for MachO.  Go ahead with your patch and I'll see what I can do with the MCJIT tests on MachO afterward.

Thanks,
Andy

-----Original Message-----
From: Tim Northover [mailto:t.p.northover at gmail.com] 
Sent: Thursday, December 13, 2012 2:12 PM
To: Kaylor, Andrew
Cc: Amara Emerson; llvm-commits
Subject: Re: [llvm-commits] [patch] Identifying code sections in MCJIT

> I thought the MCJIT regression tests were disabled for Darwin systems.  Did you do something to enable them for testing purposes?

I'm afraid not. As I said I have virtually no knowledge of MachO. My assumption was that if someone actually cared about the results they'd at least have enabled the relevant regression tests

Frankly, I think that if even the regression tests don't necessarily pass on Darwin/MCJIT then there's no grounds for blocking more subtle breakages.

Tim.

On Thu, Dec 13, 2012 at 7:09 PM, Kaylor, Andrew <andrew.kaylor at intel.com> wrote:
> I thought the MCJIT regression tests were disabled for Darwin systems.  Did you do something to enable them for testing purposes?
>
> -Andy
>
>
> -----Original Message-----
> From: Tim Northover [mailto:t.p.northover at gmail.com]
> Sent: Thursday, December 13, 2012 4:25 AM
> To: Amara Emerson
> Cc: Kaylor, Andrew; llvm-commits
> Subject: Re: [llvm-commits] [patch] Identifying code sections in MCJIT
>
>> However, I have rather limited access to MachO hardware, and even 
>> less experience with it so we could really do with an expert weighing 
>> in here. It may well be that the current implementation of isText is 
>> good enough for all supported MachO uses, or at least as good as 
>> checking the first symbol encountered in a section.
>
> Well, I have managed to dig up an OSX machine, and at least the regression tests on MachO seem to be unaffected by this change. Which suggests that if it does cause problems it's more likely to be somewhere people don't care about.
>
> Tim.




More information about the llvm-commits mailing list