gotom,
I agree with Chritophe here, even after looking at
the relevant bits in other architectures and past
comments by Ulrich.
A machine context is exactly a signal context.
That definition of mcontext_t looks extremely out
of date :}
I will test the suggested change and get back to you.
c.
Note: Due to some b0rkeness in our 64-bit kernel
the sigcontext returned may not be correct.
We are taking measures to put that fire out :)
> Hi,
>
> The forwarded mail is received, but I don't know it's correct or not
> (and I doubt the explained solution is ok).
> Could you enlighten me on this BTS?
>
> Regards,
>
> -- gotom
>
> X-Envelope-Sender: [EMAIL PROTECTED]
> Subject: Bug#157374: libc6: [parisc] sys/ucontext.h and asm/ucontext.h disagree
> Reply-To: Christophe Rhodes <[EMAIL PROTECTED]>, [EMAIL PROTECTED]
> Resent-From: Christophe Rhodes <[EMAIL PROTECTED]>
> Original-Sender: Christophe Rhodes <[EMAIL PROTECTED]>
> Resent-To: [EMAIL PROTECTED]
> Resent-Cc: GNU Libc Maintainers <[EMAIL PROTECTED]>,
> [EMAIL PROTECTED]
> Resent-Date: Tue, 20 Aug 2002 10:18:02 GMT
> Resent-Message-ID: <[EMAIL PROTECTED]>
> X-Debian-PR-Message: report 157374
> X-Debian-PR-Package: libc6
> X-Debian-PR-Keywords:
> From: Christophe Rhodes <[EMAIL PROTECTED]>
> To: [EMAIL PROTECTED]
> X-Spam-Status: No, hits=-2.6 required=4.7
>tests=FROM_ENDS_IN_NUMS,FORGED_RCVD_FOUND,SENT_BY_BTS version=2.01
> X-Mailing-List: <[EMAIL PROTECTED]> archive/latest/2576
> List-Post: <mailto:[EMAIL PROTECTED]>
> List-Help: <mailto:[EMAIL PROTECTED]?subject=help>
> List-Subscribe: <mailto:[EMAIL PROTECTED]?subject=subscribe>
> List-Unsubscribe: <mailto:[EMAIL PROTECTED]?subject=unsubscribe>
> Resent-Sender: [EMAIL PROTECTED]
>
> Package: libc6
> Version: 2.2.5-6
> Severity: normal
>
> On the parisc, sys/ucontext.h says
>
> typedef struct
> {
> gregset_t gregs;
> fpregset_t fpregs;
> } mcontext_t;
>
> which is all very well, but doesn't correspond either with asm/ucontext.h:
>
> struct ucontext {
> unsigned long uc_flags;
> struct ucontext *uc_link;
> stack_t uc_stack;
struct sigcontext uc_mcontext;
> sigset_t uc_sigmask; /* mask last for extensibility */
> };
>
> or with what the kernel puts there (which is a struct sigcontext, as far as I
> can tell from reading the kernel sources). This means that an application
> #including <sys/ucontext.h> will get the wrong layout for what's in the
> uc_mcontext slot.
>
> My preferred resolution would be to remove the typedef from sys/ucontext.h and
> replace it with
>
> typedef struct sigcontext mcontext_t;
>
> Cheers,
>
> Christophe
> -- System Information
> Debian Release: testing/unstable
> Kernel Version: Linux lambda 2.4.18-pre7-ac1 #1 Thu Jan 31 19:05:47 GMT 2002 i686
>unknown unknown GNU/Linux
>
>
>
> --
> To UNSUBSCRIBE, email to [EMAIL PROTECTED]
> with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
