Convert sigaction to act like other unices
Subsystems (9)
PATCH 1/1 Convert sigaction to act like other unices
Here's a patch that converts all architectures to behave like other unix boxes signal handling. It's funny that I didn't need to change the m68k architecture, since it was the only one that already behaves this way! (the m68knommu does not!)
Also, I noticed that not all check the return value of setting up the signal frame (not that the signal frame setup functions have return values). So this probably needs to be fixed in each architecture that doesn't already do it. Since that is a little more advanced change, I don't feel comfortable with changing that without even having the ability to compile it.
Also note, that these changes were done without compling (since I don't happen to have a compiler for every arch that Linux supports). So these are all untested, but I did go over them three times to make sure it looks good. I recommend those that support these archs to check these as well. (I tried to CC all the arch maintainers, but I might have missed some).
I did this against 2.6.13-rc6-git4, but it is best to throw this into -mm or early 2.6.14-rc's.
-- Steve
arch/alpha/kernel/signal.c +5 -6
@@ -566,13 +566,12 @@ handle_signal(int sig, struct k_sigactioif (ka->sa.sa_flags &SA_RESETHAND ) ka->sa.sa_handler =SIG_DFL ; -if (!(ka->sa.sa_flags &SA_NODEFER )) { -spin_lock_irq (¤t->sighand->siglock); -sigorsets (¤t->blocked,¤t->blocked,&ka->sa.sa_mask); +spin_lock_irq (¤t->sighand->siglock); +sigorsets (¤t->blocked,¤t->blocked,&ka->sa.sa_mask); +if (!(ka->sa.sa_flags &SA_NODEFER ))sigaddset (¤t->blocked,sig); -recalc_sigpending (); -spin_unlock_irq (¤t->sighand->siglock); - } +recalc_sigpending (); +spin_unlock_irq (¤t->sighand->siglock); } static inline void
arch/arm/kernel/signal.c +3 -2
@@ -658,11 +658,12 @@ handle_signal(unsigned long sig, struct/* * Block the signal if we were unsuccessful. */ -if (ret !=0 || !(ka->sa.sa_flags &SA_NODEFER )) { +if (ret !=0 ) {spin_lock_irq (&tsk->sighand->siglock);sigorsets (&tsk->blocked, &tsk->blocked, &ka->sa.sa_mask); -sigaddset (&tsk->blocked, sig); +if (!(ka->sa.sa_flags &SA_NODEFER )) +sigaddset (&tsk->blocked, sig);recalc_sigpending ();spin_unlock_irq (&tsk->sighand->siglock); }
arch/arm26/kernel/signal.c +6 -7
@@ -454,14 +454,13 @@ handle_signal(unsigned long sig, siginfoif (ka->sa.sa_flags &SA_ONESHOT ) ka->sa.sa_handler =SIG_DFL ; -if (!(ka->sa.sa_flags &SA_NODEFER )) { -spin_lock_irq (&tsk->sighand->siglock); -sigorsets (&tsk->blocked, &tsk->blocked, - &ka->sa.sa_mask); +spin_lock_irq (&tsk->sighand->siglock); +sigorsets (&tsk->blocked, &tsk->blocked, + &ka->sa.sa_mask); +if (!(ka->sa.sa_flags &SA_NODEFER ))sigaddset (&tsk->blocked, sig); -recalc_sigpending (); -spin_unlock_irq (&tsk->sighand->siglock); - } +recalc_sigpending (); +spin_unlock_irq (&tsk->sighand->siglock);return ; }
arch/cris/arch-v10/kernel/signal.c +5 -6
@@ -517,13 +517,12 @@ handle_signal(int canrestart, unsigned lif (ka->sa.sa_flags &SA_ONESHOT ) ka->sa.sa_handler =SIG_DFL ; -if (!(ka->sa.sa_flags &SA_NODEFER )) { -spin_lock_irq (¤t->sighand->siglock); -sigorsets (¤t->blocked,¤t->blocked,&ka->sa.sa_mask); +spin_lock_irq (¤t->sighand->siglock); +sigorsets (¤t->blocked,¤t->blocked,&ka->sa.sa_mask); +if (!(ka->sa.sa_flags &SA_NODEFER ))sigaddset (¤t->blocked,sig); -recalc_sigpending (); -spin_unlock_irq (¤t->sighand->siglock); - } +recalc_sigpending (); +spin_unlock_irq (¤t->sighand->siglock); }/*
arch/cris/arch-v32/kernel/signal.c +5 -6
@@ -568,13 +568,12 @@ handle_signal(int canrestart, unsigned lif (ka->sa.sa_flags &SA_ONESHOT ) ka->sa.sa_handler =SIG_DFL ; -if (!(ka->sa.sa_flags &SA_NODEFER )) { -spin_lock_irq (¤t->sighand->siglock); -sigorsets (¤t->blocked,¤t->blocked,&ka->sa.sa_mask); +spin_lock_irq (¤t->sighand->siglock); +sigorsets (¤t->blocked,¤t->blocked,&ka->sa.sa_mask); +if (!(ka->sa.sa_flags &SA_NODEFER ))sigaddset (¤t->blocked,sig); -recalc_sigpending (); -spin_unlock_irq (¤t->sighand->siglock); - } +recalc_sigpending (); +spin_unlock_irq (¤t->sighand->siglock); }/*
arch/frv/kernel/signal.c +5 -6
@@ -506,13 +506,12 @@ static void handle_signal(unsigned longelse setup_frame (sig, ka, oldset, regs); -if (!(ka->sa.sa_flags &SA_NODEFER )) { -spin_lock_irq (¤t->sighand->siglock); -sigorsets (¤t->blocked, ¤t->blocked, &ka->sa.sa_mask); +spin_lock_irq (¤t->sighand->siglock); +sigorsets (¤t->blocked, ¤t->blocked, &ka->sa.sa_mask); +if (!(ka->sa.sa_flags &SA_NODEFER ))sigaddset (¤t->blocked, sig); -recalc_sigpending (); -spin_unlock_irq (¤t->sighand->siglock); - } +recalc_sigpending (); +spin_unlock_irq (¤t->sighand->siglock); }/* end handle_signal() */ /*****************************************************************************/
arch/h8300/kernel/signal.c +5 -6
@@ -488,13 +488,12 @@ handle_signal(unsigned long sig, siginfoelse setup_frame (sig, ka, oldset, regs); -if (!(ka->sa.sa_flags &SA_NODEFER )) { -spin_lock_irq (¤t->sighand->siglock); -sigorsets (¤t->blocked,¤t->blocked,&ka->sa.sa_mask); +spin_lock_irq (¤t->sighand->siglock); +sigorsets (¤t->blocked,¤t->blocked,&ka->sa.sa_mask); +if (!(ka->sa.sa_flags &SA_NODEFER ))sigaddset (¤t->blocked,sig); -recalc_sigpending (); -spin_unlock_irq (¤t->sighand->siglock); - } +recalc_sigpending (); +spin_unlock_irq (¤t->sighand->siglock); }/*
arch/i386/kernel/signal.c +3 -2
@@ -577,10 +577,11 @@ handle_signal(unsigned long sig, siginfoelse ret =setup_frame (sig, ka, oldset, regs); -if (ret && !(ka->sa.sa_flags &SA_NODEFER )) { +if (ret) {spin_lock_irq (¤t->sighand->siglock);sigorsets (¤t->blocked,¤t->blocked,&ka->sa.sa_mask); -sigaddset (¤t->blocked,sig); +if (!(ka->sa.sa_flags &SA_NODEFER )) +sigaddset (¤t->blocked,sig);recalc_sigpending ();spin_unlock_irq (¤t->sighand->siglock); }
arch/ia64/kernel/signal.c +6 -9
@@ -467,15 +467,12 @@ handle_signal (unsigned long sig, structif (!setup_frame (sig, ka, info, oldset, scr))return 0 ; -if (!(ka->sa.sa_flags &SA_NODEFER )) { -spin_lock_irq (¤t->sighand->siglock); - { -sigorsets (¤t->blocked, ¤t->blocked, &ka->sa.sa_mask); -sigaddset (¤t->blocked, sig); -recalc_sigpending (); - } -spin_unlock_irq (¤t->sighand->siglock); - } +spin_lock_irq (¤t->sighand->siglock); +sigorsets (¤t->blocked, ¤t->blocked, &ka->sa.sa_mask); +if (!(ka->sa.sa_flags &SA_NODEFER )) +sigaddset (¤t->blocked, sig); +recalc_sigpending (); +spin_unlock_irq (¤t->sighand->siglock);return 1 ; }
arch/m32r/kernel/signal.c +5 -6
@@ -341,13 +341,12 @@ handle_signal(unsigned long sig, struct/* Set up the stack frame */ setup_rt_frame (sig, ka, info, oldset, regs); -if (!(ka->sa.sa_flags &SA_NODEFER )) { -spin_lock_irq (¤t->sighand->siglock); -sigorsets (¤t->blocked,¤t->blocked,&ka->sa.sa_mask); +spin_lock_irq (¤t->sighand->siglock); +sigorsets (¤t->blocked,¤t->blocked,&ka->sa.sa_mask); +if (!(ka->sa.sa_flags &SA_NODEFER ))sigaddset (¤t->blocked,sig); -recalc_sigpending (); -spin_unlock_irq (¤t->sighand->siglock); - } +recalc_sigpending (); +spin_unlock_irq (¤t->sighand->siglock); }/*
arch/m68knommu/kernel/signal.c +5 -6
@@ -732,13 +732,12 @@ handle_signal(int sig, struct k_sigactioif (ka->sa.sa_flags &SA_ONESHOT ) ka->sa.sa_handler =SIG_DFL ; -if (!(ka->sa.sa_flags &SA_NODEFER )) { -spin_lock_irq (¤t->sighand->siglock); -sigorsets (¤t->blocked,¤t->blocked,&ka->sa.sa_mask); +spin_lock_irq (¤t->sighand->siglock); +sigorsets (¤t->blocked,¤t->blocked,&ka->sa.sa_mask); +if (!(ka->sa.sa_flags &SA_NODEFER ))sigaddset (¤t->blocked,sig); -recalc_sigpending (); -spin_unlock_irq (¤t->sighand->siglock); - } +recalc_sigpending (); +spin_unlock_irq (¤t->sighand->siglock); }/*
arch/mips/kernel/irixsig.c +5 -6
@@ -155,13 +155,12 @@ static inline void handle_signal(unsigneelse setup_irix_frame (ka, regs, sig, oldset); -if (!(ka->sa.sa_flags &SA_NODEFER )) { -spin_lock_irq (¤t->sighand->siglock); -sigorsets (¤t->blocked,¤t->blocked,&ka->sa.sa_mask); +spin_lock_irq (¤t->sighand->siglock); +sigorsets (¤t->blocked,¤t->blocked,&ka->sa.sa_mask); +if (!(ka->sa.sa_flags &SA_NODEFER ))sigaddset (¤t->blocked,sig); -recalc_sigpending (); -spin_unlock_irq (¤t->sighand->siglock); - } +recalc_sigpending (); +spin_unlock_irq (¤t->sighand->siglock); } asmlinkage int do_irix_signal(sigset_t *oldset, struct pt_regs *regs)
arch/mips/kernel/signal32.c +5 -6
@@ -751,13 +751,12 @@ static inline void handle_signal(unsigneelse setup_frame (ka, regs, sig, oldset); -if (!(ka->sa.sa_flags &SA_NODEFER )) { -spin_lock_irq (¤t->sighand->siglock); -sigorsets (¤t->blocked,¤t->blocked,&ka->sa.sa_mask); +spin_lock_irq (¤t->sighand->siglock); +sigorsets (¤t->blocked,¤t->blocked,&ka->sa.sa_mask); +if (!(ka->sa.sa_flags &SA_NODEFER ))sigaddset (¤t->blocked,sig); -recalc_sigpending (); -spin_unlock_irq (¤t->sighand->siglock); - } +recalc_sigpending (); +spin_unlock_irq (¤t->sighand->siglock); }int do_signal32 (sigset_t *oldset,struct pt_regs *regs)
arch/mips/kernel/signal.c +5 -6
@@ -425,13 +425,12 @@ static inline void handle_signal(unsignesetup_frame (ka, regs, sig, oldset);#endif -if (!(ka->sa.sa_flags &SA_NODEFER )) { -spin_lock_irq (¤t->sighand->siglock); -sigorsets (¤t->blocked,¤t->blocked,&ka->sa.sa_mask); +spin_lock_irq (¤t->sighand->siglock); +sigorsets (¤t->blocked,¤t->blocked,&ka->sa.sa_mask); +if (!(ka->sa.sa_flags &SA_NODEFER ))sigaddset (¤t->blocked,sig); -recalc_sigpending (); -spin_unlock_irq (¤t->sighand->siglock); - } +recalc_sigpending (); +spin_unlock_irq (¤t->sighand->siglock); }extern int do_signal32 (sigset_t *oldset,struct pt_regs *regs);
arch/parisc/kernel/signal.c +5 -6
@@ -517,13 +517,12 @@ handle_signal(unsigned long sig, siginfoif (!setup_rt_frame (sig, ka, info, oldset, regs, in_syscall))return 0 ; -if (!(ka->sa.sa_flags &SA_NODEFER )) { -spin_lock_irq (¤t->sighand->siglock); -sigorsets (¤t->blocked,¤t->blocked,&ka->sa.sa_mask); +spin_lock_irq (¤t->sighand->siglock); +sigorsets (¤t->blocked,¤t->blocked,&ka->sa.sa_mask); +if (!(ka->sa.sa_flags &SA_NODEFER ))sigaddset (¤t->blocked,sig); -recalc_sigpending (); -spin_unlock_irq (¤t->sighand->siglock); - } +recalc_sigpending (); +spin_unlock_irq (¤t->sighand->siglock);return 1 ; }
arch/ppc/kernel/signal.c +5 -6
@@ -759,13 +759,12 @@ int do_signal(sigset_t *oldset, struct pelse handle_signal (signr, &ka, &info, oldset, regs, newsp); -if (!(ka.sa.sa_flags &SA_NODEFER )) { -spin_lock_irq (¤t->sighand->siglock); -sigorsets (¤t->blocked,¤t->blocked,&ka.sa.sa_mask); +spin_lock_irq (¤t->sighand->siglock); +sigorsets (¤t->blocked,¤t->blocked,&ka.sa.sa_mask); +if (!(ka.sa.sa_flags &SA_NODEFER ))sigaddset (¤t->blocked, signr); -recalc_sigpending (); -spin_unlock_irq (¤t->sighand->siglock); - } +recalc_sigpending (); +spin_unlock_irq (¤t->sighand->siglock);return 1 ; }
arch/ppc64/kernel/signal32.c +3 -2
@@ -976,11 +976,12 @@ int do_signal32(sigset_t *oldset, structelse ret =handle_signal32 (signr, &ka, &info, oldset, regs, newsp); -if (ret && !(ka.sa.sa_flags &SA_NODEFER )) { +if (ret) {spin_lock_irq (¤t->sighand->siglock);sigorsets (¤t->blocked, ¤t->blocked, &ka.sa.sa_mask); -sigaddset (¤t->blocked, signr); +if (!(ka.sa.sa_flags &SA_NODEFER )) +sigaddset (¤t->blocked, signr);recalc_sigpending ();spin_unlock_irq (¤t->sighand->siglock); }
arch/ppc64/kernel/signal.c +3 -2
@@ -481,10 +481,11 @@ static int handle_signal(unsigned long s/* Set up Signal Frame */ ret =setup_rt_frame (sig, ka, info, oldset, regs); -if (ret && !(ka->sa.sa_flags &SA_NODEFER )) { +if (ret) {spin_lock_irq (¤t->sighand->siglock);sigorsets (¤t->blocked, ¤t->blocked, &ka->sa.sa_mask); -sigaddset (¤t->blocked,sig); +if (!(ka->sa.sa_flags &SA_NODEFER )) +sigaddset (¤t->blocked,sig);recalc_sigpending ();spin_unlock_irq (¤t->sighand->siglock); }
arch/s390/kernel/compat_signal.c +5 -6
@@ -637,12 +637,11 @@ handle_signal32(unsigned long sig, strucelse setup_frame32 (sig, ka, oldset, regs); -if (!(ka->sa.sa_flags &SA_NODEFER )) { -spin_lock_irq (¤t->sighand->siglock); -sigorsets (¤t->blocked,¤t->blocked,&ka->sa.sa_mask); +spin_lock_irq (¤t->sighand->siglock); +sigorsets (¤t->blocked,¤t->blocked,&ka->sa.sa_mask); +if (!(ka->sa.sa_flags &SA_NODEFER ))sigaddset (¤t->blocked,sig); -recalc_sigpending (); -spin_unlock_irq (¤t->sighand->siglock); - } +recalc_sigpending (); +spin_unlock_irq (¤t->sighand->siglock); }
arch/s390/kernel/signal.c +5 -6
@@ -429,13 +429,12 @@ handle_signal(unsigned long sig, structelse setup_frame (sig, ka, oldset, regs); -if (!(ka->sa.sa_flags &SA_NODEFER )) { -spin_lock_irq (¤t->sighand->siglock); -sigorsets (¤t->blocked,¤t->blocked,&ka->sa.sa_mask); +spin_lock_irq (¤t->sighand->siglock); +sigorsets (¤t->blocked,¤t->blocked,&ka->sa.sa_mask); +if (!(ka->sa.sa_flags &SA_NODEFER ))sigaddset (¤t->blocked,sig); -recalc_sigpending (); -spin_unlock_irq (¤t->sighand->siglock); - } +recalc_sigpending (); +spin_unlock_irq (¤t->sighand->siglock); }/*
arch/sh/kernel/signal.c +5 -6
@@ -546,13 +546,12 @@ handle_signal(unsigned long sig, structif (ka->sa.sa_flags &SA_ONESHOT ) ka->sa.sa_handler =SIG_DFL ; -if (!(ka->sa.sa_flags &SA_NODEFER )) { -spin_lock_irq (¤t->sighand->siglock); -sigorsets (¤t->blocked,¤t->blocked,&ka->sa.sa_mask); +spin_lock_irq (¤t->sighand->siglock); +sigorsets (¤t->blocked,¤t->blocked,&ka->sa.sa_mask); +if (!(ka->sa.sa_flags &SA_NODEFER ))sigaddset (¤t->blocked,sig); -recalc_sigpending (); -spin_unlock_irq (¤t->sighand->siglock); - } +recalc_sigpending (); +spin_unlock_irq (¤t->sighand->siglock); }/*
arch/sh64/kernel/signal.c +5 -6
@@ -664,13 +664,12 @@ handle_signal(unsigned long sig, siginfoelse setup_frame (sig, ka, oldset, regs); -if (!(ka->sa.sa_flags &SA_NODEFER )) { -spin_lock_irq (¤t->sighand->siglock); -sigorsets (¤t->blocked,¤t->blocked,&ka->sa.sa_mask); +spin_lock_irq (¤t->sighand->siglock); +sigorsets (¤t->blocked,¤t->blocked,&ka->sa.sa_mask); +if (!(ka->sa.sa_flags &SA_NODEFER ))sigaddset (¤t->blocked,sig); -recalc_sigpending (); -spin_unlock_irq (¤t->sighand->siglock); - } +recalc_sigpending (); +spin_unlock_irq (¤t->sighand->siglock); }/*
arch/sparc/kernel/signal.c +5 -6
@@ -1034,13 +1034,12 @@ handle_signal(unsigned long signr, strucelse setup_frame (&ka->sa, regs, signr, oldset, info); } -if (!(ka->sa.sa_flags &SA_NOMASK )) { -spin_lock_irq (¤t->sighand->siglock); -sigorsets (¤t->blocked,¤t->blocked,&ka->sa.sa_mask); +spin_lock_irq (¤t->sighand->siglock); +sigorsets (¤t->blocked,¤t->blocked,&ka->sa.sa_mask); +if (!(ka->sa.sa_flags &SA_NOMASK ))sigaddset (¤t->blocked, signr); -recalc_sigpending (); -spin_unlock_irq (¤t->sighand->siglock); - } +recalc_sigpending (); +spin_unlock_irq (¤t->sighand->siglock); } static inline void syscall_restart(unsigned long orig_i0, struct pt_regs *regs,
arch/sparc64/kernel/signal32.c +5 -6
@@ -1325,13 +1325,12 @@ static inline void handle_signal32(unsigelse setup_frame32 (&ka->sa, regs, signr, oldset, info); } -if (!(ka->sa.sa_flags &SA_NOMASK )) { -spin_lock_irq (¤t->sighand->siglock); -sigorsets (¤t->blocked,¤t->blocked,&ka->sa.sa_mask); +spin_lock_irq (¤t->sighand->siglock); +sigorsets (¤t->blocked,¤t->blocked,&ka->sa.sa_mask); +if (!(ka->sa.sa_flags &SA_NOMASK ))sigaddset (¤t->blocked,signr); -recalc_sigpending (); -spin_unlock_irq (¤t->sighand->siglock); - } +recalc_sigpending (); +spin_unlock_irq (¤t->sighand->siglock); } static inline void syscall_restart32(unsigned long orig_i0, struct pt_regs *regs,
arch/sparc64/kernel/signal.c +5 -6
@@ -574,13 +574,12 @@ static inline void handle_signal(unsigne {setup_rt_frame (ka, regs, signr, oldset, (ka->sa.sa_flags &SA_SIGINFO ) ? info :NULL ); -if (!(ka->sa.sa_flags &SA_NOMASK )) { -spin_lock_irq (¤t->sighand->siglock); -sigorsets (¤t->blocked,¤t->blocked,&ka->sa.sa_mask); +spin_lock_irq (¤t->sighand->siglock); +sigorsets (¤t->blocked,¤t->blocked,&ka->sa.sa_mask); +if (!(ka->sa.sa_flags &SA_NOMASK ))sigaddset (¤t->blocked,signr); -recalc_sigpending (); -spin_unlock_irq (¤t->sighand->siglock); - } +recalc_sigpending (); +spin_unlock_irq (¤t->sighand->siglock); } static inline void syscall_restart(unsigned long orig_i0, struct pt_regs *regs,
arch/um/kernel/signal_kern.c +3 -3
@@ -87,12 +87,12 @@ static int handle_signal(struct pt_regsrecalc_sigpending ();spin_unlock_irq (¤t->sighand->siglock);force_sigsegv (signr, current); - } -else if (!(ka->sa.sa_flags &SA_NODEFER )){ + }else {spin_lock_irq (¤t->sighand->siglock);sigorsets (¤t->blocked, ¤t->blocked, &ka->sa.sa_mask); -sigaddset (¤t->blocked, signr); +if (!(ka->sa.sa_flags &SA_NODEFER )) +sigaddset (¤t->blocked, signr);recalc_sigpending ();spin_unlock_irq (¤t->sighand->siglock); }
arch/v850/kernel/signal.c +5 -6
@@ -462,13 +462,12 @@ handle_signal(unsigned long sig, siginfoelse setup_frame (sig, ka, oldset, regs); -if (!(ka->sa.sa_flags &SA_NODEFER )) { -spin_lock_irq (¤t->sighand->siglock); -sigorsets (¤t->blocked,¤t->blocked,&ka->sa.sa_mask); +spin_lock_irq (¤t->sighand->siglock); +sigorsets (¤t->blocked,¤t->blocked,&ka->sa.sa_mask); +if (!(ka->sa.sa_flags &SA_NODEFER ))sigaddset (¤t->blocked,sig); -recalc_sigpending (); -spin_unlock_irq (¤t->sighand->siglock); - } +recalc_sigpending (); +spin_unlock_irq (¤t->sighand->siglock); }/*
arch/x86_64/kernel/signal.c +3 -2
@@ -394,10 +394,11 @@ handle_signal(unsigned long sig, siginfo#endif ret =setup_rt_frame (sig, ka, info, oldset, regs); -if (ret && !(ka->sa.sa_flags &SA_NODEFER )) { +if (ret) {spin_lock_irq (¤t->sighand->siglock);sigorsets (¤t->blocked,¤t->blocked,&ka->sa.sa_mask); -sigaddset (¤t->blocked,sig); +if (!(ka->sa.sa_flags &SA_NODEFER )) +sigaddset (¤t->blocked,sig);recalc_sigpending ();spin_unlock_irq (¤t->sighand->siglock); }
arch/xtensa/kernel/signal.c +5 -6
@@ -702,12 +702,11 @@ int do_signal(struct pt_regs *regs, sigsif (ka.sa.sa_flags &SA_ONESHOT ) ka.sa.sa_handler =SIG_DFL ; -if (!(ka.sa.sa_flags &SA_NODEFER )) { -spin_lock_irq (¤t->sighand->siglock); -sigorsets (¤t->blocked, ¤t->blocked, &ka.sa.sa_mask); +spin_lock_irq (¤t->sighand->siglock); +sigorsets (¤t->blocked, ¤t->blocked, &ka.sa.sa_mask); +if (!(ka.sa.sa_flags &SA_NODEFER ))sigaddset (¤t->blocked, signr); -recalc_sigpending (); -spin_unlock_irq (¤t->sighand->siglock); - } +recalc_sigpending (); +spin_unlock_irq (¤t->sighand->siglock);return 1 ; }
On Fri, Aug 12, 2005 at 10:40:02PM -0400, Steven Rostedt wroqte:
Quoted 4 lines
Here's a patch that converts all architectures to behave like other unix boxes signal handling. It's funny that I didn't need to change the m68k architecture, since it was the only one that already behaves this way! (the m68knommu does not!)
<rest snipped which also wasn't better>
This is not a description of what you changed. A patch entry has to start with a rationale and then a description of the change.
-Andi
Quoted 10 lines · Andi Kleen
On Fri, Aug 12, 2005 at 10:40:02PM -0400, Steven Rostedt wroqte:
Quoted 4 lines
Here's a patch that converts all architectures to behave like other unix boxes signal handling. It's funny that I didn't need to change the m68k architecture, since it was the only one that already behaves this way! (the m68knommu does not!)
<rest snipped which also wasn't better>
This is not a description of what you changed. A patch entry has to start with a rationale and then a description of the change.
Sorry, I forgot that not all were in on the thread. (duh, I added a bunch of others, I guess I wasn't thinking clearly).
http://lkml.org/lkml/2005/8/9/190 "Singal handling possibly wrong".
Here's a summary:
Bodo Stroesser noticed a conflict between the man pages and what the kernel was doing in regard to SA_NODEFER. The man pages stated that SA_NODEFER would only not defer the signal that is being processed, where as the the kernel would not defer all signals (including those in sa_mask).
The POSIX description of this was very confusing, where Linus thought it explained what the kernel currently does, and I (and others) thought it did what the man pages described. Someone found an updated version that supports my theory. In fact it seems that even the signal would be defered if SA_NODEFER was set but the signal was in the sa_mask.
Finally, it was ask, what do other unix boxes do? I wrote a simple program that tested the behavior of this and posted it. I had a couple of responses that tested the following unices:
AIX, DU4, IRIX6, Solaris and OSF1, and all of these do it the way the man pages described. Since Linux happens to be the odd ball out, I submitted the patch to make Linux act like other unix boxes with regard to signals and SA_NODEFER.
Hope this explains things better.
-- Steve
Quoted 7 lines · Steven Rostedt
Quoted 6 lines · Andi Kleen
On Fri, Aug 12, 2005 at 10:40:02PM -0400, Steven Rostedt wroqte:
Quoted 4 lines
Here's a patch that converts all architectures to behave like other unix boxes signal handling. It's funny that I didn't need to change the m68k architecture, since it was the only one that already behaves this way! (the m68knommu does not!)
Quoted 7 lines
Quoted 4 lines
<rest snipped which also wasn't better>
This is not a description of what you changed. A patch entry has to start with a rationale and then a description of the change.
Sorry, I forgot that not all were in on the thread. (duh, I added a bunch of others, I guess I wasn't thinking clearly).
I actually read the thread, but patches with useless descriptions like your previous one cannot be applied in general because you cannot expect that people later looking at the <VCS system of the week> logs know about some obscure thread months or years before. Or rather the person who applied it would need to write a better description on his own, which is not their job.
My general feeling about the change is that it risks breaking programs and doesn't seem to have any compelling advantages,
so unless there is a bug demonstrated I wouldn't apply it.
-Andi
Quoted 8 lines · Andi Kleen
Quoted 7 lines · Steven Rostedt
Quoted 6 lines · Andi Kleen
On Fri, Aug 12, 2005 at 10:40:02PM -0400, Steven Rostedt wroqte:
Quoted 4 lines
Here's a patch that converts all architectures to behave like other unix boxes signal handling. It's funny that I didn't need to change the m68k architecture, since it was the only one that already behaves this way! (the m68knommu does not!)
[snip]
Quoted 7 lines
My general feeling about the change is that it risks breaking programs and doesn't seem to have any compelling advantages,
so unless there is a bug demonstrated I wouldn't apply it.-Andi
As I see it, the advantages are that we would a) match the documentation (man pages & posix/SUS) which makes things easier for application writers who won't have to scratch their beards wondering why Linux doesn't behave like the docs say. And b) Linux behaviour would match what most (all?) other Unices do, so there'll be less
hassle/bugs when porting apps from other systems to Linux. To me,
those look like significant bennefits.
As for the "it may break programs" bit, that of course is a concern, but one way around that would be to stick it in -mm and let it cook for a few kernel releases. Say we stick it in -mm with a plan to merge it into 2.6.16-rc1, that should give it quite a bit of time to determine if it breaks apps (and if it does, to fix those apps).
Just my 0.02euro.
Signature · 3 lines
Jesper Juhl <[email protected]> Don't top-post http://www.catb.org/~esr/jargon/html/T/top-post.html Plain text mails only, please http://www.expita.com/nomime.html
Quoted 18 lines · Jesper Juhl
Quoted 8 lines · Andi Kleen
Quoted 7 lines · Steven Rostedt
Quoted 6 lines · Andi Kleen
On Fri, Aug 12, 2005 at 10:40:02PM -0400, Steven Rostedt wroqte:
Quoted 4 lines
Here's a patch that converts all architectures to behave like other unix boxes signal handling. It's funny that I didn't need to change the m68k architecture, since it was the only one that already behaves this way! (the m68knommu does not!)
[snip]
Quoted 7 lines
My general feeling about the change is that it risks breaking programs and doesn't seem to have any compelling advantages,
so unless there is a bug demonstrated I wouldn't apply it.-Andi
Quoted 12 lines
As I see it, the advantages are that we would a) match the documentation (man pages & posix/SUS) which makes things easier for application writers who won't have to scratch their beards wondering why Linux doesn't behave like the docs say. And b) Linux behaviour would match what most (all?) other Unices do, so there'll be less
hassle/bugs when porting apps from other systems to Linux. To me,
those look like significant bennefits.
As for the "it may break programs" bit, that of course is a concern, but one way around that would be to stick it in -mm and let it cook for a few kernel releases. Say we stick it in -mm with a plan to merge it into 2.6.16-rc1, that should give it quite a bit of time to determine if it breaks apps (and if it does, to fix those apps).
Has there been a verdict on this. Are we going to
a) leave Linux "as-is" and be different than the man pages and POSIX as well as pretty much every other Unix out there. But..., we keep from breaking any linux-only applications out there that actually depend on this behaviour (if there were any).
b) add the patch (in -mm or early 14 or later), see if any applications break, but we will finally match the man pages and etc.
Just curious.
-- Steve
Quoted 3 lines · Steven Rostedt
b) add the patch (in -mm or early 14 or later), see if any applications break, but we will finally match the man pages and etc.
We'll definitely test the patch. I doubt it breaks anything, and it's the right thing to do, but yes, we'll try it out very early in the 2.6.14 cycle. Somebody make sure to send me the patch after I release 2.6.13 so that I don't forget.
Linus
Quoted 1 line · Andi Kleen
<rest snipped which also wasn't better>
Also, what you snipped was another issue, not addressed by the patch. Some architectures test if the setting up of the signal handler stack frame succeeded, and others don't. I believe that this may be a problem, since it was recently fixed on the x86, and probably should be addressed on the architectures that don't check for failure.
-- Steve