Den 9. jul. 2010 21.02 skrev Gary Thomas <[email protected]>: > Sorry, I misinterpreted the dump - it's in the CLD code. Issue filed. >
Glad to hear it. For the record V8 is supposed to work on machines/OSs that don't allow misaligned accesses, but there's a build-time define if your setup allows it. Performance (esp. regexp) is better with unaligned accesses. > > On Jul 9, 12:14 pm, Gary Thomas <[email protected]> wrote: > > I'm running Chrome (6.0.451.0/r50986) on an embedded ARM > > (OMAP3530==armv7a) > > system. Periodically, I get a slew of these errors while pages are > > being rendered: > > Alignment trap: chrome (3185) PC=0x00d7ae08 Instr=0xe8961002 > > Address=0x02606377 FSR 0x001 > > > > Looking at the memory map for chrome, it's obvious this is in DATA: > > # cat /proc/3223/maps > > 00008000-02471000 r-xp 00000000 b3:02 82835 /usr/bin/chrome/ > > chrome > > 02478000-02489000 rwxp 02468000 b3:02 82835 /usr/bin/chrome/ > > chrome > > 02489000-026ea000 rwxp 00000000 00:00 0 [heap] > > > > I'm guessing that this is in V8 generated code (it's not in any code > > segment of any of the threads) > > > > Any ideas about this? The page seems to render properly - I think the > > kernel makes the appropriate data movements to make things work, but > > of course having to take a trap into the kernel on such activity > > really slows things down. > > > > Thanks for any pointers > > -- > v8-dev mailing list > [email protected] > http://groups.google.com/group/v8-dev > -- Erik Corry, Software Engineer Google Denmark ApS. CVR nr. 28 86 69 84 c/o Philip & Partners, 7 Vognmagergade, P.O. Box 2227, DK-1018 Copenhagen K, Denmark. -- v8-dev mailing list [email protected] http://groups.google.com/group/v8-dev
