Hi,
thanks for catching that. I really didn't want to push this change, it was
part of my debugging. The problem is that when CDFS crashes at this point,
VFS hangs, so taskdump won't be run. The kernel will print register
contents, but you don't know the address space map so you can't determine,
whether the address being accessed is valid or not.
In cases where the system is still able to run taskdump, you get this
information. The address space map would be useful not only for amd64+ia32,
but for any architecture. If we wanted that, it might be useful implementing
a common handler for bad memory access across architectures.
Cheers,
Jiri
---------- Původní e-mail ----------
Od: Jakub Jermář <[email protected]>
Komu: [email protected]
Datum: 9. 7. 2017 22:18:07
Předmět: Re: [HelenOS-devel] [HelenOS-commits] [mainline] r2689 - Fix CDFS
unaligned memory access when decoding Joliet file names.
"Hi Jiri,
I noticed this change:
On 07/08/2017 06:12 PM, [email protected] wrote:
> Repo: mainline
> Committer: Jiri Svoboda <jiri@wiwaxia>
> Date: 2017-07-08 18:12:03 CEST (Sat, 08 Jul 2017)
> New Revision: 2689
> New Id: jiri@wiwaxia-20170708161203-ti7p0ahv4bx63v3x
> Parent: jiri@wiwaxia-20170706161919-sa6kbgpacvmwrhb7
>
> Log:
> Fix CDFS unaligned memory access when decoding Joliet file names.
>
> Modified:
> kernel/arch/amd64/src/interrupt.c
> uspace/srv/fs/cdfs/cdfs_ops.c
>
> === modified file 'kernel/arch/amd64/src/interrupt.c'
> --- kernel/arch/amd64/src/interrupt.c 2017-06-19 21:47:42 +0000
> +++ kernel/arch/amd64/src/interrupt.c 2017-07-08 16:12:03 +0000
> @@ -134,6 +134,7 @@
> io_perm_bitmap_install();
> return;
> }
> + as_print(TASK->as);
> fault_if_from_uspace(istate, "General protection fault.");
> }
> panic_badtrap(istate, n, "General protection fault.");
and was wondering if it is just a left-over from your debugging session
or whether you really wanted to add this to the handling of the GPF. Or
in other words, what useful information did this give you compared to
the mere fault_if_from_uspace(). Can this be useful also on ia32 or for
other types of faults?
Thanks,
Jakub
_______________________________________________
HelenOS-devel mailing list
[email protected]
http://lists.modry.cz/listinfo/helenos-devel
"
_______________________________________________
HelenOS-devel mailing list
[email protected]
http://lists.modry.cz/listinfo/helenos-devel
