d/freebsd-skq
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

The per-CPU spinlocks list is only maintained when WITNESS is enabled.

Browse Source 
Thus, treat all page faults while in a critical section as fatal rather
than just those that occur with a non-empty spinlocks list.  All such page
faults are fatal anyways.  Calling trap_fatal() earlier increases the
chances of getting more useful panic messages and a possible DDB prompt.

Approved by:	re (scottl)
This commit is contained in:
jhb 2003-05-20 20:50:33 +00:00
parent 94215f2da0
commit a808871f56
1 changed files with 5 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
sys/i386/i386/trap.c
View File

@ -222,10 +222,10 @@ trap(frame)
type);
/*
* Page faults need interrupts diasabled until later,
* and we shouldn't enable interrupts while holding a
* spin lock.
* and we shouldn't enable interrupts while in a
* critical section.
*/
if (type != T_PAGEFLT && PCPU_GET(spinlocks) == NULL)
if (type != T_PAGEFLT && td->td_critnest == 0)
enable_intr();
}
}
@ -240,7 +240,7 @@ trap(frame)
* are finally ready to read %cr2 and then must
* reenable interrupts.
*
* If we get a page fault while holding a spin lock, then
* If we get a page fault while in a critical section, then
* it is most likely a fatal kernel page fault. The kernel
* is already going to panic trying to get a sleep lock to
* do the VM lookup, so just consider it a fatal trap so the
@ -248,7 +248,7 @@ trap(frame)
* to the debugger.
*/
eva = rcr2();
if (PCPU_GET(spinlocks) == NULL)
if (td->td_critnest == 0)
enable_intr();
else
trap_fatal(&frame, eva);