Messages in this thread | | | From | Andi Kleen <> | Subject | Re: [PATCH] x86 NUMA panic compile error | Date | Mon, 15 May 2006 22:02:30 +0200 |
| |
On Monday 15 May 2006 21:59, Andrew Morton wrote:
> > On Monday 15 May 2006 21:39, Andrew Morton wrote: > > > Ingo Molnar <mingo@elte.hu> wrote: > > > > > > > > Nevertheless for hard-to-debug bugs i prefer if they can be reproduced > > > > and debugged on 32-bit too, because x86_64 debugging is still quite a > > > > PITA and wastes alot of time: for example it has no support for exact > > > > kernel stacktraces. Also, the printout of the backtrace is butt-ugly and > > > > as un-ergonomic to the human eye as it gets > > > > > > Yes, I find x86_64 traces significantly harder to follow. And I miss the > > > display of the length of the functions (do_md_run+1208 instead of > > > do_md_run+1208/2043). The latter form makes it easier to work out > > > whereabouts in the function things happened. > > > > > > That, plus the mix of hex and decimal numbers.. > > > > > > > who came up with that > > > > "two-maybe-one function entries per-line" nonsense? [Whoever did it he > > > > never had to look at (and make sense of) hundreds of stacktraces in a > > > > row.] > > > > > > Plus they're wide enough to get usefully wordwrapped when someone mails > > > them to you. > > > > Hmm, I didn't realize they were _that_ unpopular. If you got the i386 > > like space wasting backtraces would you guys all switch your development machines > > to x86-64 ? @) > > > > Developers use serial consoles for such things. (I discovered > `console=uart,...' yesterday. It works nicely as an earlyprintk on ia64..)
I can also recommend firescope + firewire cards. It's not early yet, but I hope eventually. But it can work without the target still being alive and also does on most laptops.
> It's reports-from-the-field which are the problem.
In my experience the biggest problem in the field is that most of it scrolls away. That is why I tweaked the x86-64 format to be as space efficient as possible. That's also why the "executive summary" was added.
But Ingo has a point that it usually doesn't help anyways because backtraces tend to be so overlong now after the code got through 20 callbacks before it can do something actually useful.
> A lot of these problems can be address by simple cranking up the VGA screen > resolution, but I discovered that I don't know how to do that - I've always > used `vga=extended', but that doesn't work on an EFI-booted ia64 box. > > Does anyone know what the magic option is to make the vga console use 50 > rows?
I use vga=0x0f07
It's a butt ugly font, but it's the smallest I could find without using the slow fbcon.
If you can't remember the hex number use vga=ask
-Andi
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
| |