This is the mail archive of the gdb-patches@sources.redhat.com mailing list for the GDB project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH] auxv support


> Here's the board file I use to run locally, gross hacks and all.

I can't figure out where to put this file and how to tell runtest to use it.

> > However, on a system that does not support getting the auxv data, it shows
> > three failures.  What is the right thing to do about this?  The difficulty
> > is that the error from `info auxv' does not distinguish an error/bug in
> > reading the data from the target code not supporting auxv access or from
> > the native system not supporting the access even though the gdb target code
> > does (e.g. Linux < 2.6).
> 
> Perhaps arrange to distinguish in the error message?

This sounds like it should be easy, but it's really not.  The
target-config-never-supports case and error cases are both indicated by
target_read_partial returning -1.  No other details are made available by
the target interfaces.  Even if we addressed the case of there not being
any code around that does it at all, there are still the kernel (or stub)
doesn't support it vs miscellaneous error cases.  The backend or gdbserver
code can potentially distinguish, i.e. failing open with ENOENT vs other
errors or errors on read after open.  But that would require passing back
the failure mode in some distinguished way, rather than just -1.


Thanks,
Roland


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]