Re: [PATCH 4/4] 'catch syscall' feature -- Build system, documentation and testcase

[Sérgio Durigan Júnior <sergiodj at linux dot vnet dot ibm dot com>]

Hi Tom,

On Tue, 2008-11-04 at 09:32 -0700, Tom Tromey wrote:
> >>>>> "Sérgio" == Sérgio Durigan Júnior <sergiodj@linux.vnet.ibm.com> writes:
> 
> Sérgio> This last part modifies the build system (in order to
> Sérgio> implement the GDB's datadir), adds the documentation and the
> Sérgio> testcase for the new feature.
> 
> Nice.
> 
> Sérgio> +xml-syscall-copy:
> Sérgio> +	list='$(XML_SYSCALLS_FILES)' ; \
> Sérgio> +	for file in $$list ; do \
> 
> Two nits here.
> 
> First, if you build with builddir!=srcdir, you need to mkdir
> XML_SYSCALLS_DIR here.
> 
> Second, I think it would be nice to detect the case where
> builddir==srcdir and not copy in that case.

You're right. Thanks, I'll fix it.

> Sérgio> +xml-syscall-install:
> Sérgio> +	$(SHELL) $(srcdir)/../mkinstalldirs \
> Sérgio> +		$(DESTDIR)$(GDB_DATADIR_PATH)/$(XML_SYSCALLS_DIR) ; \
> 
> Using DESTDIR is nice attention to detail :-)

Haha, I'm learning :-)

> Sérgio> +# Also, the "install" target installs the syscalls' XML files in the system.
> Sérgio> +install: all install-only xml-syscall-install
> 
> I think install-only should depend on xml-syscall-install.
> I believe the intent here is that "make install-only" will install
> everything without rebuilding.

Ok, consider it done.

> Sérgio> +AC_ARG_WITH(gdb-datadir,
> Sérgio> +[  --with-gdb-datadir=path   Look for global separate data files in this path [DATADIR/gdb]],
> 
> It is preferable, IMO, to use AS_HELP_STRING here.
> There are some examples of this in gdb's configure.ac.

OK, no problem by me. But I did a cut & paste from the other parts
(specifically the DEBUGDIR part), which does exactly this :-). Just to
let you know.

> Sérgio> +AC_CONFIG_SUBDIRS(doc testsuite syscalls)
>  
> You don't want this.  There is no syscalls directory, and even if
> there was, I would argue that it should not have its own configure
> script.

Sure, this was totally my fault. I'm sorry.

> Sérgio> +@item syscall
> Sérgio> +@itemx syscall @var{syscall name}
> Sérgio> +@cindex break on call to/return from a system call
> Sérgio> +A call to/return from a @code{syscall} -- similar to the @code{strace}
> Sérgio> +utility.
> 
> This needs a minor update now that "catch syscall" can take multiple
> arguments.
> 
> I didn't read the implementation patch in detail... if it parses
> numeric syscalls, then that should be mentioned here.

Yeah, thanks. Eli already pointed this error too. Thanks.

> 
> Finally, instead of "call to/return from", how about "call to or
> return from"?  This is more grammatical and not much longer :-)
> 
> Tom
-- 
Sérgio Durigan Júnior
Linux on Power Toolchain - Software Engineer
Linux Technology Center - LTC
IBM Brazil