| T.R | Title | User | Personal
Name | Date | Lines |
|---|
| 92.1 | Hope this helps | LANDO::CUMMINS | | Thu Jun 06 1996 10:18 | 71 |
| 92.2 | No undetected references to memory holes. | POBOXA::DUNCAN | | Thu Jun 06 1996 17:33 | 11 |
| 92.3 | thanks and more questions | EWBV05::TAKAHASHI_HI | Technical Support/MCS/Japan | Thu Jun 06 1996 23:52 | 36 |
| 92.4 | Reply .2 is incorrect. Further explanation.. | LANDO::CUMMINS | | Fri Jun 07 1996 12:04 | 56 |
| 92.5 | Answers to questions in .3 | LANDO::CUMMINS | | Fri Jun 07 1996 12:15 | 46 |
| 92.6 | Wrap | EWBV05::TAKAHASHI_HI | Technical Support/MCS/Japan | Mon Jun 10 1996 04:18 | 19 |
| 92.7 | | LANDO::CUMMINS | | Mon Jun 10 1996 11:16 | 55 |
| 92.8 | Thanks again | EWBV05::TAKAHASHI_HI | Technical Support/MCS/Japan | Mon Jun 10 1996 23:19 | 28 |
| 92.9 | | STAR::CROLL | | Tue Jun 11 1996 11:23 | 19 |
| 92.10 | | EWBV05::TAKAHASHI_HI | Technical Support/MCS/Japan | Tue Jun 11 1996 22:05 | 4 |
| 92.11 | 2.5GB=3.5GB! | TMCUKA::ROWELL | Paul Rowell @BBP | Fri Mar 21 1997 06:18 | 8 |
| I have just had the situation where I installed 1GB and 3 * 512MB
options and then booted VMS 6.2-1h3.
The operating system then thinks it has 3.5GB, with disastrous results.
I consider this to be rather alarming!!!
-Paul
|
| 92.12 | Need info from you | HARMNY::CUMMINS | | Fri Mar 21 1997 09:43 | 17 |
| The SRM console initializes, tests, and configures main memory and then
passes memory descriptors and bitmaps to UNIX, VMS, and AlphaBIOS/NT at
boot time to describe the amount of available memory and the number of
good/bad/tested pages. Could you please post the following information?
P00>>> show version
P00>>> show memory
P00>>> b -h ...
.
.
P00>>> info 1
.
.
P00>>>
Thanks,
BC
|
| 92.13 | Information requested. | TMCUKA::ROWELL | Paul Rowell @BBP | Fri Mar 21 1997 11:46 | 92 |
| Ok, the customers have left early so I can put the machine back how it
was when we had the problem...
P00>>>sh version
version V3.0-10, 19-NOV-1996 13:57:07
P00>>>sh mem
Slot Type MB Base
---- ---- ---- --------
0 EDO 1024 0
1 EDO 512 40000000
2 EDO 512 80000000
3 EDO 512 c0000000
Total 2560
P00>>>b -h
(boot dke201.2.0.3.0 -flags 10,0)
Building FRU table
block 0 of dke201.2.0.3.0 is a valid boot block
reading 1004 blocks from dke201.2.0.3.0
bootstrap code read in
base = 200000, image_start = 0, image_bytes = 7d800
initializing HWRPB at 2000
initializing page table at 1f2000
initializing machine state
setting affinity to the primary CPU
P00>>>info 1
HWRPB: 2000 MEMDSC:2d80 Cluster count: 5
CLUSTER: 0 Usage: Console
BITMAP_VA: n/a BITMAP_PA: n/a START_PFN: 0
PFN_COUNT:100 TESTED_COUNT: 0
CLUSTER: 1 Usage: System
BITMAP_VA: 101ef800 BITMAP_PA: dfff5800 START_PFN: 100
PFN_COUNT:2ff00 TESTED_COUNT: 2ff00
GOOD from page 100:2ffff (address range 0000000000200000:000000005fffe000)
CLUSTER: 2 Usage: System
BITMAP_VA: 101f57e0 BITMAP_PA: dfffb7e0 START_PFN: 40000
PFN_COUNT:10000 TESTED_COUNT: 10000
GOOD from page 40000:4ffff (address range
0000000080000000:000000009fffe000)
CLUSTER: 3 Usage: System
BITMAP_VA: 101f77e0 BITMAP_PA: dfffd7e0 START_PFN: 60000
PFN_COUNT:fffa TESTED_COUNT: fffa
GOOD from page 60000:6fff9 (address range
00000000c0000000:00000000dfff2000)
CLUSTER: 4 Usage: Console
BITMAP_VA: n/a BITMAP_PA: n/a START_PFN: 6fffa
PFN_COUNT:6 TESTED_COUNT: 0
P00>>>
And just to show I wasn't making it up...
$ sh mem
System Memory Resources on 21-MAR-1997 16:34:50.47
Physical Memory Usage (pages): Total Free In Use
Modified
Main Memory (3584.00Mb) 458752 317255 141429
68
Virtual I/O Cache (Kbytes): Total Free In Use
Cache Memory 3200 0 3200
Granularity Hint Regions (pages): Total Free In Use
Released
Execlet code region 512 0 395
117
Execlet data region 128 0 88
40
VMS exec data region 4239 0 4239
0
Resident image code region 1024 213 811
0
Slot Usage (slots): Total Free Resident
Swapped
Process Entry Slots 3959 3940 19
0
Balance Set Slots 670 653 17
0
Dynamic Memory Usage (bytes): Total Free In Use
Largest
Nonpaged Dynamic Memory 19955712 13262720 6692992
13183040
Paged Dynamic Memory 23527424 21059808 2467616
21058560
Paging File Usage (blocks): Free Reservable
Total
DISK$ALPHASYS:[SYS10.SYSEXE]SWAPFILE.SYS 7936 7936
7936
DISK$ALPHASYS:[SYS10.SYSEXE]PAGEFILE.SYS 9984 1792
9984
DISK$PAGE_4K2:[PAGE]PAGEFILE.SYS;1 1999872 1949616
1999872
Of the physical pages in use, 140166 pages are permanently allocated to
OpenVMS.
$
|
| 92.14 | | HARMNY::CUMMINS | | Fri Mar 21 1997 13:14 | 12 |
| This looks like either a VMS bug or a system that has MEMORY_TEST set
less than FULL. I assume that MEMORY_TEST is set FULL on this system
since INFO 1 says tested page count equals PFN page count for those
clusters deemed usable for operating system use.
I've forwarded mail to the OpenVMS development group. Will let you know
what I hear back.
Have you by any chance run Digital UNIX on this same machine? If so,
how much memory does UNIX think it has?
BC
|
| 92.15 | Might be able to try that next week. | TMCUKA::ROWELL | Paul Rowell @BBP | Fri Mar 21 1997 13:19 | 8 |
| MEMORY_TEST is set to FULL
I haven't run UNIX on this config but if I get a chance next week when
the customers are gone and before the memory cards go back to the
rental company I'll give it a whirl, also be interesting to build V7.1
and see what that thinks?
-Paul
|
| 92.16 | Don't do that.... | LEDER1::BOWSE | | Fri Mar 21 1997 13:54 | 8 |
|
re: 11. This is expected behavior with VMS V6.2x. Memory holes
are *not* supported. V7.1 does support memory holes. If you go
back to one of Bill's earlier replies (.2), he does a good job of
explaining how memory holes happen on Rawhide. UNIX 3.2x and
subsequent support memory holes.
- Walt
|
| 92.17 | | MAY30::CUMMINS | | Fri Mar 21 1997 16:15 | 3 |
| Yes! Walt is correct. VMS V7.1 and beyond supports memory holes.
V6.2-1H3 does not. You will need to remove the last two 512MB cards
for VMS 6.2 to work properly on said machine.
|
| 92.18 | Where? | TMCUKA::ROWELL | Paul Rowell @BBP | Mon Mar 24 1997 04:19 | 4 |
| Where is this restriction mentioned, the SOC simply says that the
system supports 4 memory options and the largest must go in slot 0?
-Paul
|
| 92.19 | Physical Memory Holes- Restrictions | STAR::KFOLLIEN | | Wed Mar 26 1997 10:54 | 21 |
| Paul,
This restriction is clearly stated in the OpenVMS V6.2-1H3 Release
Notes. The stated restriction is as follows :
1.6.3 Memory Restriction
The OpenVMS Alpha Version 6.2-1H3 operating system does not
efficiently support physical memory holes. The operating
system assumes that physical memories are contiguous
and, as a result, creates certain non-pageable memory
management structures that are larger than they need to
be. This problem will be addressed in a future version of
the operating system. Until then, Digital recommends that
you avoid configuring systems in this manner.
Cheers,
Ken Follien
OpenVMS Systems Group
|