pkgsrc-Users archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: wip/chromium




To: pkgsrc-wip-discuss%lists.sourceforge.net@localhost, pkgsrc-users%netbsd.org@localhost, port-amd64%netbsd.org@localhost, port-i386%netbsd.org@localhost

Subject: Re: wip/chromium

From: Rui-Xiang Guo <rxg%lavabit.com@localhost>

Date: Thu, 20 Jan 2011 14:06:58 +0800


On Sun, Jan 16, 2011 at 03:01:32PM +0100, Joerg Sonnenberger wrote:
> On Sun, Jan 16, 2011 at 12:06:11PM +0800, Rui-Xiang Guo wrote:
> > Program received signal SIGSEGV, Segmentation fault.
> > tcmalloc::ThreadCache::GetThreadHeap ()
> >     at third_party/tcmalloc/chromium/src/thread_cache.h:375
> > 375         return threadlocal_heap_;
> 
> threadlocal_heap_ is declared as __thread by chance?

Good hint. It is controled by HAVE_TLS -
    371 inline ThreadCache* ThreadCache::GetThreadHeap() {
    372 #ifdef HAVE_TLS
    373   // __thread is faster, but only when the kernel supports it
    374   if (KernelSupportsTLS())
    375     return threadlocal_heap_;
    376 #endif
    377   return reinterpret_cast<ThreadCache *>(
    378       perftools_pthread_getspecific(heap_key_)); 
    379 }

HAVE_TLS is defined if compiler supports __thread.

I disable the option manually. It looks better now and help me to find some
missing patches.

Is it safe to close it?

-rxg


References:

wip/chromium
From: Rui-Xiang Guo

Re: wip/chromium
From: Rui-Xiang Guo

Re: wip/chromium
From: Pouya D. Tafti

Re: wip/chromium
From: Thomas Klausner

Re: wip/chromium
From: Rui-Xiang Guo

Re: wip/chromium
From: Joerg Sonnenberger




Prev by Date: Re: pkgtools/osabi from pkgsrc-2010Q4 fails to install on aix 5.3

Next by Date: Re: [HEADS UP] png update to 1.5 branch

Previous by Thread: Re: wip/chromium

Next by Thread: chromium 10

Indexes:

reverse Date

reverse Thread

Old Index



Home | Main Index | Thread Index | Old Index