2008-03-17 09:21:52

by Roland McGrath

Subject: [PATCH] x86 handle_vm86_trap cleanup

Use force_sig in handle_vm86_trap like other machine traps do.

Signed-off-by: Roland McGrath <[email protected]>
---
arch/x86/kernel/vm86_32.c | 9 +--------
1 files changed, 1 insertions(+), 8 deletions(-)

diff --git a/arch/x86/kernel/vm86_32.c b/arch/x86/kernel/vm86_32.c
index 738c210..fddb998 100644
--- a/arch/x86/kernel/vm86_32.c
+++ b/arch/x86/kernel/vm86_32.c
@@ -553,16 +553,9 @@ int handle_vm86_trap(struct kernel_vm86_regs * regs, long error_code, int trapno
}
if (trapno !=1)
return 1; /* we let this handle by the calling routine */
- if (current->ptrace & PT_PTRACED) {
- unsigned long flags;
- spin_lock_irqsave(&current->sighand->siglock, flags);
- sigdelset(&current->blocked, SIGTRAP);
- recalc_sigpending();
- spin_unlock_irqrestore(&current->sighand->siglock, flags);
- }
- send_sig(SIGTRAP, current, 1);
current->thread.trap_no = trapno;
current->thread.error_code = error_code;
+ force_sig(SIGTRAP, current);
return 0;
}

2008-03-21 10:40:15

by Ingo Molnar

Subject: Re: [PATCH] x86 handle_vm86_trap cleanup

* Roland McGrath <[email protected]> wrote:

> Use force_sig in handle_vm86_trap like other machine traps do.

thanks, applied.

> - if (current->ptrace & PT_PTRACED) {
> - unsigned long flags;
> - spin_lock_irqsave(&current->sighand->siglock, flags);
> - sigdelset(&current->blocked, SIGTRAP);
> - recalc_sigpending();
> - spin_unlock_irqrestore(&current->sighand->siglock, flags);
> - }
> - send_sig(SIGTRAP, current, 1);
> current->thread.trap_no = trapno;
> current->thread.error_code = error_code;
> + force_sig(SIGTRAP, current);
> return 0;
> }

for the record, this also changes semantics, because previously we'd
permanently unblock SIGTRAP (most certainly as a hack, to be able to do
the send_sig()), while now we use force_sig() which just ignores the
blocked mask. (and i agree with your fix of course)

Ingo

2008-03-22 03:54:44

by Roland McGrath

Subject: Re: [PATCH] x86 handle_vm86_trap cleanup

> for the record, this also changes semantics, because previously we'd
> permanently unblock SIGTRAP (most certainly as a hack, to be able to do
> the send_sig()), while now we use force_sig() which just ignores the
> blocked mask. (and i agree with your fix of course)

That's not so. force_sig_info (called by force_sig) removes the signal
from the blocked set too. That is the only way to get the signal ever to
be dequeued.


Thanks,
Roland

2008-03-22 11:17:52

by Ingo Molnar

Subject: Re: [PATCH] x86 handle_vm86_trap cleanup

* Roland McGrath <[email protected]> wrote:

> > for the record, this also changes semantics, because previously we'd
> > permanently unblock SIGTRAP (most certainly as a hack, to be able to
> > do the send_sig()), while now we use force_sig() which just ignores
> > the blocked mask. (and i agree with your fix of course)
>
> That's not so. force_sig_info (called by force_sig) removes the
> signal from the blocked set too. That is the only way to get the
> signal ever to be dequeued.

ah, indeed, i missed that :)

Ingo