Re: port-xen/58561 (panic: kernel diagnostic assertion, "x86_read_psl() == 0" failed: file, "/home/netbsd/10/src/sys/arch/x86/x86/pmap.c", line 3581)

["Manuel Bouyer via gnats" <gnats-admin%NetBSD.org@localhost>]

The following reply was made to PR port-xen/58561; it has been noted by GNATS.

From: Manuel Bouyer <bouyer%antioche.eu.org@localhost>
To: Konrad Schroder <perseant%hhhh.org@localhost>
Cc: gnats-bugs%NetBSD.org@localhost, port-xen-maintainer%netbsd.org@localhost,
        netbsd-bugs%netbsd.org@localhost, gnats-admin%netbsd.org@localhost, riastradh%NetBSD.org@localhost,
        campbell+netbsd%mumble.net@localhost, cherry%NetBSD.org@localhost
Subject: Re: port-xen/58561 (panic: kernel diagnostic assertion,
 "x86_read_psl() == 0" failed: file,
 "/home/netbsd/10/src/sys/arch/x86/x86/pmap.c", line 3581)
Date: Sat, 10 Jan 2026 22:01:05 +0100

 --L07qoGXM5nk0oetQ
 Content-Type: text/plain; charset=iso-8859-1
 Content-Disposition: inline
 Content-Transfer-Encoding: 8bit
 
 On Sat, Jan 10, 2026 at 11:30:27AM -0800, Konrad Schroder wrote:
 > On 1/10/2026 4:55 AM, Manuel Bouyer wrote:
 > > Hello,
 > > can you try with the attached patch ? It won't fix the problem but
 > > should let us know if syscall() is already called with interrupts disabled,
 > > or if they're disabled later
 > 
 > Thanks!  Unfortunately that blows up almost instantly:
 > [...]
 
 
 Sorry; I should have checked more carefully. As there's a
 call _C_LABEL(do_pmap_load)
 early I assumed it was safe to call C functions here but do_pmap_load()
 is in fact written in assembly.
 Here's an updated patch which uses only assembly in copy.S
 Also it adds the check to all copy* functions, not only copyout.
 It boots multiuser on my test system.
 
 -- 
 Manuel Bouyer <bouyer%antioche.eu.org@localhost>
      NetBSD: 26 ans d'experience feront toujours la difference
 --
 
 --L07qoGXM5nk0oetQ
 Content-Type: text/plain; charset=us-ascii
 Content-Disposition: attachment; filename=diff
 
 Index: sys/arch/amd64/amd64/copy.S
 ===================================================================
 RCS file: /cvsroot/src/sys/arch/amd64/amd64/copy.S,v
 retrieving revision 1.36
 diff -u -p -u -r1.36 copy.S
 --- sys/arch/amd64/amd64/copy.S	24 Sep 2024 20:19:30 -0000	1.36
 +++ sys/arch/amd64/amd64/copy.S	10 Jan 2026 20:56:54 -0000
 @@ -47,12 +47,14 @@
  	movq	CPUVAR(CURLWP),reg; \
  	movq	L_PCB(reg),reg
  
 +
  /*
   * These are arranged so that the abnormal case is a forwards
   * conditional branch - which will be predicted not-taken by
   * both Intel and AMD processors.
   */
  #define DEFERRED_SWITCH_CHECK \
 +	call _C_LABEL(do_check_psl)		; \
  	CHECK_DEFERRED_SWITCH			; \
  	jnz	99f				; \
  98:
 @@ -70,6 +72,33 @@
  
  x86_copyfunc_start:	.globl	x86_copyfunc_start
  
 +pslmsg:	.ascii "do_check_psl failed"
 +ENTRY(do_check_psl)
 +	pushq	%rbp
 +	movq	%rsp,%rbp
 +	pushq	%rbx
 +	movq	CPUVAR(CURLWP),%rbx
 +	incl    L_NOPREEMPT(%rbx)
 +	movq	CPUVAR(VCPU), %rax
 +	movzbl EVTCHN_UPCALL_MASK(%rax),%eax
 +	decl	L_NOPREEMPT(%rbx)
 +	test   %eax,%eax
 +	jne 	198f
 +	popq	%rbx
 +	leaveq
 +	ret
 +198:
 +	movq $pslmsg, %rdi
 +	callq _C_LABEL(panic)
 +	popq	%rbx
 +	leaveq
 +	ret
 +END(do_check_psl)
 +
 +
 +
 +
 +
  /*
   * Handle deferred pmap switch.  We must re-enable preemption without
   * making a function call, so that the program counter is visible to
 Index: sys/arch/x86/x86/syscall.c
 ===================================================================
 RCS file: /cvsroot/src/sys/arch/x86/x86/syscall.c,v
 retrieving revision 1.22
 diff -u -p -u -r1.22 syscall.c
 --- sys/arch/x86/x86/syscall.c	5 Oct 2023 19:41:06 -0000	1.22
 +++ sys/arch/x86/x86/syscall.c	10 Jan 2026 20:56:54 -0000
 @@ -48,6 +48,7 @@ __KERNEL_RCSID(0, "$NetBSD: syscall.c,v 
  #include <machine/userret.h>
  
  #include "opt_dtrace.h"
 +#include "opt_xen.h"
  
  #ifndef __x86_64__
  int		x86_copyargs(void *, void *, size_t);
 @@ -90,6 +91,10 @@ static
  void
  syscall(struct trapframe *frame)
  {
 +#ifdef XENPV
 +		/* Check to see if interrupts are enabled (ie; no events are masked) */
 +		KASSERT(x86_read_psl() == 0);
 +#endif
  	const struct sysent *callp;
  	struct proc *p;
  	struct lwp *l;
 
 --L07qoGXM5nk0oetQ--