shortcuts in dmar_gas matching

[dougm]

Summary

In dmar_gas_lowermatch, skip searching a subtree if all its addresses are greater than lowaddr.

In dmar_gas_uppermatch, skip searching a subtree if all its gaps-between-alloctions are too small.

[pho]

I probably won't be able to test this before tomorrow.

[pho]

With "hw.dmar.enable=1" I see:

20200129 15:36:54 all (1/1): udp.sh
Expensive timeout(9) function: 0xffffffff80c5bbb0(0) 0.016073906 s
Expensive timeout(9) function: 0xffffffff80be2300(0xfffff80846234530) 0.024061505 s
Expensive timeout(9) function: 0xffffffff80c5bc80(0) 0.081290860 s
Expensive timeout(9) function: 0xffffffff80c5bc80(0) 0.231948206 s
(da0:isci0:0:0:0): WRITE(10). CDB: 2a 00 07 7d 6b 62 00 00 40 00 
(da0:isci0:0:0:0): CAM status: CCB request completed with an error
(da0:isci0:0:0:0): Retrying command, 3 more tries remain
swap_pager: indefinite wait buffer: bufobj: 0, blkno: 1580935, size: 16384
isci: 1580309000:401318 ISCI Sending reset to device on controller 0 domain 0 CAM index 0
isci: 1580309000:410015 ISCI isci: bus=0 target=0 lun=0 cdb[0]=615539f0 terminated
isci: 1580309000:418124 ISCI isci: bus=0 target=0 lun=0 cdb[0]=5c5e9f0 terminated
isci: 1580309000:426163 ISCI isci: bus=0 target=0 lun=0 cdb[0]=3eae1f0 terminated
isci: 1580309000:434210 ISCI isci: bus=0 target=0 lun=0 cdb[0]=615671f0 terminated
isci: 1580309000:442363 ISCI isci: bus=0 target=0 lun=0 cdb[0]=5c461f0 terminated
isci: 1580309000:450402 ISCI isci: bus=0 target=0 lun=0 cdb[0]=41c0a9f0 terminated
isci: 1580309000:458552 ISCI isci: bus=0 target=0 lun=0 cdb[0]=5c471f0 terminated
isci: 1580309000:466601 ISCI isci: bus=0 target=0 lun=0 cdb[0]=6d50b1f0 terminated
isci: 1580309000:474741 ISCI isci: bus=0 target=0 lun=0 cdb[0]=6d50b9f0 terminated
(da0:isci0:0:0:0): WRITE(10). CDB: 2a 00 3e 25 fc aa 00 01 00 00 
(da0:isci0:0:0:0): CAM status: CCB request terminated by the host
(da0:isci0:0:0:0): Retrying command, 3 more tries remain

[dougm]

The lines added here were also in https://reviews.freebsd.org/D23189?id=66970 which passed this test, so I'm stumped at the moment.

[pho]

After trying a bit harder with r357254 I see the same problem, so it is unrelated to this patch.

[dougm]

Before I commit this patch, can you determine if the problem was present before or after r357173?

[pho]

I only have one UDP test that have triggered this.
On r357172 this test ran for 5 hours without any problems.
On r357173 I saw the problem after running the test for 30 minutes.

[dougm]

Subsequent testing by Peter appears to have absolved r357173 of blame, so I'm going ahead with this commit.

[kib]

In D23391#514734, @dougm wrote:

Subsequent testing by Peter appears to have absolved r357173 of blame, so I'm going ahead with this commit.

Do you mean that isci(4) was broken anyway ? Are any other drivers were reported as broken ?

[dougm]

In D23391#514739, @kib wrote:

In D23391#514734, @dougm wrote:

Subsequent testing by Peter appears to have absolved r357173 of blame, so I'm going ahead with this commit.

Do you mean that isci(4) was broken anyway ? Are any other drivers were reported as broken ?

You can see exactly what Peter said at https://reviews.freebsd.org/D23435.

[pho]

In D23391#514739, @kib wrote:

In D23391#514734, @dougm wrote:

Subsequent testing by Peter appears to have absolved r357173 of blame, so I'm going ahead with this commit.

Do you mean that isci(4) was broken anyway ? Are any other drivers were reported as broken ?

I'm currently double checking this, by once again doing a commit search to see where the problem was introduced.
I only have one test: udp.sh which triggers the problem, and run time to reproduce the problem varies a lot.

[pho]

I have now managed to trigger the problem in r357172, so this IMHO definitely exonerates r357273.

(da0:isci0:0:0:0): SCSI status: Check Condition
(da0:isci0:0:0:0): SCSI sense: UNIT ATTENTION asc:29,2 (SCSI bus reset occurred)
(da0:isci0:0:0:0): Field Replaceable Unit: 2
(da0:isci0:0:0:0): Retrying command (per sense data)
Feb  2 18:06:37 t2 kernel: pid 7384 (swap), jid 0, uid 1004, was killed: out of swap space

FreeBSD/amd64 (t2.osted.lan) (ttyu1)

login: root
Password:
Last login: Sun Feb  2 14:36:27 on ttyu1
FreeBSD 13.0-CURRENT (PHO) #0 r357172M: Sun Feb  2 17:54:56 CET 2020
:
:
root@t2:~ # uname -a
FreeBSD t2.osted.lan 13.0-CURRENT FreeBSD 13.0-CURRENT #0 r357172M: Sun Feb  2 17:54:56 CET 2020     [e-mail]:/usr/src/sys/amd64/compile/PHO  amd64
root@t2:~ # svnlite diff /usr/src/sys
Index: /usr/src/sys/vm/vm_fault.c
===================================================================
--- /usr/src/sys/vm/vm_fault.c  (revision 357172)
+++ /usr/src/sys/vm/vm_fault.c  (working copy)
@@ -1073,12 +1073,14 @@
                    fs->oom < vm_pfault_oom_attempts) {
                        fs->oom++;
                        vm_waitpfault(dset, vm_pfault_oom_wait * hz);
+               } else  {
+                       if (bootverbose)
+                               printf(
+               "proc %d (%s) failed to alloc page on fault, starting OOM\n",
+                                   curproc->p_pid, curproc->p_comm);
+                       vm_pageout_oom(VM_OOM_MEM_PF);
+                       fs->oom = 0;
                }
-               if (bootverbose)
-                       printf(
-"proc %d (%s) failed to alloc page on fault, starting OOM\n",
-                           curproc->p_pid, curproc->p_comm);
-               vm_pageout_oom(VM_OOM_MEM_PF);
                return (KERN_RESOURCE_SHORTAGE);
        }
        fs->oom = 0;
root@t2:~ # 
$

I have only observed problems with isci.