This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


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

Re: egcs-19980508 powerpc-ibm-aix4.1.4.0 results (failures with -p/-pg)


 > From: David Edelsohn <dje@watson.ibm.com>
 > Subject: Re: egcs-19980508 powerpc-ibm-aix4.1.4.0 results (failures with -p/-pg) 
 > Date: Fri, 22 May 1998 19:00:00 -0400
 > 
 > 	I haven't had a chance to look at it in detail, but I don't think
 > the -p/-pg problem with GCC nested functions is as simple as Mike assumes.
 > I think that the __mcount calls actually are being done correctly but that
 > __mcount may not interact well with nested functions, e.g. __mcount may
 > blow out the stack given the way the stack is created for nested functions
 > and AIX __mcount's use of the stack.
 > 
 > 	The failure only occurs if mcrt0.o is used as the start file --
 > simply inserting __mcount calls is not the problem.  Some of the failures
 > are KILL signal and its hard to understand what a program could do which
 > would prompt the OS to hand it that signal -- it frequently occurs when a
 > program eats up all memory.
 > 
 > David

	Thanks for the info David.  I agree that spending time on a bug
in a gcc C language extension which occurs only when using -p/-pg on AIX
seems like a waste of time.  However, I wonder if this bug impacts other
front ends which make use of this feature. 

	Eg, don't Fortran and Pascal make use of gcc's nested function
support?  If so, would this bug mean that standards conforming Fortran or
Pascal code couldn't be profiled using g77 or gpc?  I can't say, but I
thought I'd bring it up in case you could clear it up. 

		Thanks,
		--Kaveh
--
Kaveh R. Ghazi			Project Manager / Custom Development
ghazi@caip.rutgers.edu		Icon CMT Corp.


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