Fix handling of NMIs from unknown sources (BMC, hypervisor)

Release kernels have no KDB backends enabled, so they discard an NMI
if it is not due to a hardware failure.  This includes NMIs from
IPMI BMCs and hypervisors.

Furthermore, the interaction of panic_on_nmi, kdb_on_nmi, and
debugger_on_panic is confusing.

Respond to all NMIs according to panic_on_nmi and debugger_on_panic.
Remove kdb_on_nmi.  Expand the meaning of panic_on_nmi by making
it a bitfield.  There are currently two bits: one for NMIs due to
hardware failure, and one for all others.  Leave room for more.

If panic_on_nmi and debugger_on_panic are both true, don't actually panic,
but directly enter the debugger, to allow someone to leave the debugger
and [hopefully] resume normal execution.

Reviewed by:	kib
MFC after:	2 weeks
Relnotes:	yes: machdep.kdb_on_nmi is gone; machdep.panic_on_nmi changed
Sponsored by:	Dell EMC Isilon
Differential Revision:	https://reviews.freebsd.org/D24558
This commit is contained in:
vangyzen 2020-04-26 00:41:29 +00:00
parent bccd2db598
commit 3188fdb7b8
3 changed files with 24 additions and 17 deletions

View File

@ -119,9 +119,9 @@ SYSCTL_INT(_kern, OID_AUTO, panic_reboot_wait_time, CTLFLAG_RWTUN,
#ifdef KDB
#ifdef KDB_UNATTENDED
static int debugger_on_panic = 0;
int debugger_on_panic = 0;
#else
static int debugger_on_panic = 1;
int debugger_on_panic = 1;
#endif
SYSCTL_INT(_debug, OID_AUTO, debugger_on_panic,
CTLFLAG_RWTUN | CTLFLAG_SECURE,

View File

@ -65,6 +65,7 @@ struct kdb_dbbe {
SET_DECLARE(kdb_dbbe_set, struct kdb_dbbe);
extern u_char kdb_active; /* Non-zero while in debugger. */
extern int debugger_on_panic; /* enter the debugger on panic. */
extern int debugger_on_trap; /* enter the debugger on trap. */
extern struct kdb_dbbe *kdb_dbbe; /* Default debugger backend or NULL. */
extern struct trapframe *kdb_frame; /* Frame to kdb_trap(). */

View File

@ -823,20 +823,14 @@ cpu_idle_tun(void *unused __unused)
}
SYSINIT(cpu_idle_tun, SI_SUB_CPU, SI_ORDER_MIDDLE, cpu_idle_tun, NULL);
static int panic_on_nmi = 1;
static int panic_on_nmi = 0xff;
SYSCTL_INT(_machdep, OID_AUTO, panic_on_nmi, CTLFLAG_RWTUN,
&panic_on_nmi, 0,
"Panic on NMI raised by hardware failure");
"Panic on NMI: 1 = H/W failure; 2 = unknown; 0xff = all");
int nmi_is_broadcast = 1;
SYSCTL_INT(_machdep, OID_AUTO, nmi_is_broadcast, CTLFLAG_RWTUN,
&nmi_is_broadcast, 0,
"Chipset NMI is broadcast");
#ifdef KDB
int kdb_on_nmi = 1;
SYSCTL_INT(_machdep, OID_AUTO, kdb_on_nmi, CTLFLAG_RWTUN,
&kdb_on_nmi, 0,
"Go to KDB on NMI with unknown source");
#endif
void
nmi_call_kdb(u_int cpu, u_int type, struct trapframe *frame)
@ -847,19 +841,31 @@ nmi_call_kdb(u_int cpu, u_int type, struct trapframe *frame)
/* machine/parity/power fail/"kitchen sink" faults */
if (isa_nmi(frame->tf_err)) {
claimed = true;
if (panic_on_nmi)
if ((panic_on_nmi & 1) != 0)
panic("NMI indicates hardware failure");
}
#endif /* DEV_ISA */
/*
* NMIs can be useful for debugging. They can be hooked up to a
* pushbutton, usually on an ISA, PCI, or PCIe card. They can also be
* generated by an IPMI BMC, either manually or in response to a
* watchdog timeout. For example, see the "power diag" command in
* ports/sysutils/ipmitool. They can also be generated by a
* hypervisor; see "bhyvectl --inject-nmi".
*/
#ifdef KDB
if (!claimed && kdb_on_nmi) {
/*
* NMI can be hooked up to a pushbutton for debugging.
*/
printf("NMI/cpu%d ... going to debugger\n", cpu);
kdb_trap(type, 0, frame);
if (!claimed && (panic_on_nmi & 2) != 0) {
if (debugger_on_panic) {
printf("NMI/cpu%d ... going to debugger\n", cpu);
claimed = kdb_trap(type, 0, frame);
}
}
#endif /* KDB */
if (!claimed && panic_on_nmi != 0)
panic("NMI");
}
void