>>>>> "Joseph" == Joseph S Myers <jos...@codesourcery.com> writes:

Tom> This patch adds the plugin to the gcc tree and updates the
Tom> top-level configury.

Following up on your review.

Joseph> I don't see anything obvious that would disable the plugin if
Joseph> plugins are unsupported (e.g. on Windows host) or disabled
Joseph> (--disable-plugin).  Probably the relevant support from
Joseph> gcc/configure.ac needs to go somewhere it can be used at
Joseph> toplevel.

I've moved some relevant code to a new .m4 file in config and used it
from the plugin itself.  This seemed simpler than dealing with it at the
top level.  The plugin also self-disables if its configury needs are not
met.

Tom> +  self->args.push_back ("gcc");

Joseph> seems wrong - at least you should use the appropriate compiler
Joseph> name after transformation for cross compilers /
Joseph> --program-transform-name.  Though really the *versioned* driver
Joseph> $(target_noncanonical)-gcc-$(version) is the right one to use,

This turned out to be a pain :-) There are two basic problems.

First, gdb gets the names of its architectures from BFD, which doesn't
always use the same naming scheme as the GNU configury triplets.  It
does generally use the same names, but for x86 targets it differs quite
a bit.

Second, the configury triplets can vary in annoying ways that don't
really affect correct operation.  For example, "i586-" versus "i686-"
(there is a difference, but I think ignorable given the compiler flags
in the debuginfo, and anyway I suspect not discoverable by gdb); or
"-unknown-" versus "-pc-" (completely irrelevant AFAIK); or even
"x86_64-redhat-linux" versus "x86_64-unknown-linux-gnu" (seemingly
gratuitous).

In the end I added some code to gdb and to libcc1.so to construct a
regexp matching plausible results and then search $PATH for matches.
Which seems rather gross, but workable in reasonable scenarios.

I didn't try to apply the program transform name.  I suppose I could
apply it to the final "gcc" component of the name, though, without much
trouble.  I'll fix this up tomorrow.

Let me know if you have any issue with the above.  Barring that, I will
be resubmitting the series soon, most likely this week.

thanks,
Tom

Reply via email to