Resent-Date: Wed, 6 Oct 1999 08:47:51 +0200 (MET DST) Sender: labr@ubi.se Date: Wed, 06 Oct 1999 08:47:45 +0200 From: lars brinkhoff X-Accept-Language: en To: Roman Zippel CC: linux-m68k@lists.linux-m68k.org Subject: Re: PATCH: syscall change patch, part II References: Resent-From: linux-m68k@phil.uni-sb.de Roman Zippel wrote: > On Tue, 5 Oct 1999, lars brinkhoff wrote: > > - if (addr == PT_ORIG_D0) > > + if (addr == PT_ORIG_D0 && > > + (data >= NR_syscalls || data < 0)) > > goto out; > Sorry, that I have to complain again :), but IMO that's the wrong > behaviour, it should be possible to set PT_ORIG_D0 to any value and > instead the traced syscall should fail with ENOSYS. Perfectly fine with me, as long as I'm allowed to change PT_ORIG_D0 at all! In that case, I suggest this patch: --- linux-2.3.16/arch/m68k/kernel/ptrace.c.orig Tue Oct 5 14:20:38 1999 +++ linux-2.3.16/arch/m68k/kernel/ptrace.c Wed Oct 6 08:39:10 1999 @@ -416,8 +416,6 @@ addr = addr >> 2; /* temporary hack. */ - if (addr == PT_ORIG_D0) - goto out; if (addr == PT_SR) { data &= SR_MASK; data <<= 16; --- linux-2.3.16/arch/m68k/kernel/entry.S.orig Mon Oct 4 15:33:19 1999 +++ linux-2.3.16/arch/m68k/kernel/entry.S Wed Oct 6 08:45:51 1999 @@ -93,9 +93,13 @@ jbsr SYMBOL_NAME(syscall_trace) RESTORE_SWITCH_STACK addql #4,%sp + movel #-ENOSYS,PT_D0(%sp) + movel PT_ORIG_D0(%sp),%d2 + cmpl #NR_syscalls,%d2 + jcc 1f jbsr @(SYMBOL_NAME(sys_call_table),%d2:l:4)@(0) movel %d0,%sp@(PT_D0) | save the return value - subql #4,%sp | dummy return address +1: subql #4,%sp | dummy return address SAVE_SWITCH_STACK jbsr SYMBOL_NAME(syscall_trace)