[Gc] Absurd stack bottom value on ARM

Boehm, Hans hans.boehm at hp.com
Mon Mar 16 15:14:31 PST 2009


I think this assertion used a fairly arbitrary value in order to catch cases in which we misunderstood the /proc/self/stat format or the like.  Since we now have a reason to be less arbitrary, I'd suggest chopping two zeroes off the constant in the test.  A value of less than a megabyte would be really suspicious, since it wouldn't give us much room for a downward-growing stack, and we have reasons to believe that the only platform with upward-growing stacks (PA-RISC) doesn't do this either.

I'll apply this patch to the CVS tree.

Hans

> -----Original Message-----
> From: gc-bounces at napali.hpl.hp.com 
> [mailto:gc-bounces at napali.hpl.hp.com] On Behalf Of Michael Starzinger
> Sent: Saturday, March 14, 2009 2:39 PM
> To: gc at napali.hpl.hp.com
> Subject: [Gc] Absurd stack bottom value on ARM
> 
> Hello!
> 
> As you might be aware, we are using the Boehm GC in CACAO [1] 
> at the moment. Recently we received a bug report about a 
> failing assertion on an ARM board (Openmoko Neo Freerunner) 
> running the Openmoko firmware.  
> Please take a look at our Bugzilla entry [2] for a detailed 
> description of the problem.
> 
> Currently I don't have access to an Openmoko installation, so 
> I'm merely forwarding the information I received. The bugging 
> assertion [3] is still present in the recent 7.1 version of 
> the Boehm GC, so I am assuming that it hasn't been addressed yet.
> 
> Is it OK to simply remove the bugging assertion for our purposes?
> Does the "absurd stack bottom value" have any other side effects?
> 
> References:
> [1] http://www.cacaovm.org/
> [2] 
> http://server.complang.tuwien.ac.at/cgi-bin/bugzilla/show_bug.
> cgi?id=121
> [3] 
> http://mips.complang.tuwien.ac.at/hg/cacao/file/f162d02fffb3/s
> rc/mm/boehm-gc/os_dep.c#l1080
> 
> -michi
> 
> _______________________________________________
> Gc mailing list
> Gc at linux.hpl.hp.com
> http://www.hpl.hp.com/hosted/linux/mail-archives/gc/
> 


More information about the Gc mailing list