headers
der Mouse | 26 Jul 08:40

envstat wrong: who's at fault?

I've got an i386 laptop on which envstat is printing out inconsistent
figures, and I'm wondering who's at fault (and, in particular, whether
it's something software can fix).

Specifically, here are the "acpibat0 discharge rate" and "acpibat0
charge" lines from several consecutive samples taken at one-minute
intervals:

acpibat0 discharge rate:     14.220 W
        acpibat0 charge:     47.320 Wh (90.20%)

acpibat0 discharge rate:     14.240 W
        acpibat0 charge:     47.060 Wh (89.71%)

acpibat0 discharge rate:     14.260 W
        acpibat0 charge:     46.740 Wh (89.10%)

acpibat0 discharge rate:     14.290 W
        acpibat0 charge:     46.410 Wh (88.47%)

acpibat0 discharge rate:     14.290 W
        acpibat0 charge:     46.110 Wh (87.90%)

acpibat0 discharge rate:     14.310 W
        acpibat0 charge:     45.850 Wh (87.40%)

The problem is that the charge is decreasing significantly faster than
the discharge rate indicates.  If the charge figures are accurate, the
discharge rate here is more like 18-19 W (eg, consider the third and
fourth data points above: (46.74 Wh - 46.41 Wh) / (1/60 h) = 19.8 W).
(Continue reading)

Permalink | Reply | 10 comments |
headers
Christoph Egger | 25 Jul 00:31
Picon
Picon

rdtscp instruction


Hi!

Both AMD and Intel CPUs support the RDTSCP instruction.

'cpuctl indentify 0'

prints it in the features3 line.

The differences between RDTSC and RDTSCP are:

- RDTSCP is a serializing instruction
- RDTSCP additionally loads the lower 32bits
  of MSR_TSC_AUX into ecx register.

The MSR_TSC_AUX (0xc0000103) is read-write
for the OS and can be filled with anything
meaningful to the OS (i.e. processor ID).
The upper 32bits are reserved.
The MSR exists for each CPU-core.

The question is: What useful information
can NetBSD store into MSR_TSC_AUX ?

Christoph
Permalink | Reply | 6 comments |
headers
Aaron J. Grier | 16 Jul 03:16

raw vs block speed difference?

I've got 5.1_RC3 on a dell poweredge 2650 I've been eyeing for a backup
server:

ahc1 at pci3 dev 6 function 0: Adaptec aic7899 Ultra160 SCSI adapter
ahc1: interrupting at ioapic1 pin 14
ahc1: aic7899: Ultra160 Wide Channel A, SCSI Id=7, 32/253 SCBs
scsibus0 at ahc1: 16 targets, 8 luns per target
scsibus0: waiting 2 seconds for devices to settle...
sd0 at scsibus0 target 0 lun 0: <SEAGATE, ST373453LC, 0006> disk fixed
sd0: 70007 MB, 31310 cyl, 8 head, 572 sec, 512 bytes/sect x 143374744 sectors
sd0: sync (12.50ns offset 63), 16-bit (160.000MB/s) transfers, tagged queueing

rsd0a: 1342177280 bytes transferred in 17.588 secs (76312103 bytes/sec)
sd0a : 1342177280 bytes transferred in 40.806 secs (32891664 bytes/sec)

why is the raw device > twice the speed of the block device?

--

-- 
  Aaron J. Grier | "Not your ordinary poofy goof." | agrier <at> poofygoof.com
Permalink | Reply | 4 comments |
headers
Jean-Yves Migeon | 11 Jul 01:44
Picon
Favicon

PAE and balloon benchmarks

Hi lists,

FWIW, I just uploaded a wiki page for my results regarding PAE
benchmarks [1].

In short:

- expect 10-20% overhead in specific benchmarks for PAE; nothing new
here, there have been critics on PAE performance loss multiple times.

- yes, having no MP capable kernel for Xen makes a big different if you
launch parallel builds within the same domain :)

- balloon thread added by cherry@ does not seem to introduce regression
for process/thread creation and scheduling.

I have (more) detailed results if anybody wants them.

In essence, unless you need some functionality only offered by PAE (NX
bit - when supported - , 4GB+ physical memory support without having 64
bits, or Xen 3.3+ compatilibty), I suppose you are better off going the
64 bits route, especially with 32 bits compat mode.

[1] http://wiki.netbsd.org/users/jym/benchmarks/

Cheers,

--

-- 
Jean-Yves Migeon
jeanyves.migeon <at> free.fr
(Continue reading)

Permalink | Reply | 4 comments |
headers
bifferos ' | 5 Jul 14:25

Any i8259/APIC experts?

I'm trying to get NetBSD up and running on an RDC321x CPU (a 486SX
SoC with built-in ethernet and USB controllers), however as soon as I
enable OHCI support the kernel no longer boots.  As part of OHCI
initialisation there seems to be initialisation of the i8259, and it's in the
function i8259_reinit_irqs(void)  that things go wrong.

outb(IO_ICU1 + PIC_OCW1, i8259_imen);   <---- system lock
outb(IO_ICU2 + PIC_OCW1, i8259_imen >> 8);

I looked into disabling the APIC, however I don't see any easy way of
doing that, playing around with the options relating to this just leave
me with compilation errors.

Any ideas?

thanks,
Biff.

PS:  Project notes are here if anyone is interested:
http://sites.google.com/site/bifferboard/Home/netbsd
Permalink | Reply | 0 comments |
headers
Ray Phillips | 21 Jun 11:25
Picon
Picon
Favicon
Gravatar

Max VM process usage

I'm running Roaring Penguin's CanIt-PRO on a NetBSD/i386 5.0.2 
machine (although it isn't a supported platform) which has 2 GB of 
RAM and 5GB of swap space.  One of it's processes frequently exits 
unexpectedly.  ktrace has revealed the probable cause:

   3716      1 perl     CALL  break(0x1e910000)
   3716      1 perl     RET   break -1 errno 12 Cannot allocate memory

and the developers have asked "what is the maximum virtual memory 
size of a process on NetBSD on your architecture?"  Would the correct 
answer be "there is no limit" based on the following sysctl output, 
or am I looking in the wrong place?

# /sbin/sysctl proc.curproc
proc.curproc.corename = %n.core
proc.curproc.rlimit.cputime.soft = unlimited
proc.curproc.rlimit.cputime.hard = unlimited
proc.curproc.rlimit.filesize.soft = unlimited
proc.curproc.rlimit.filesize.hard = unlimited
proc.curproc.rlimit.datasize.soft = 268435456
proc.curproc.rlimit.datasize.hard = 3221225472
proc.curproc.rlimit.stacksize.soft = 2097152
proc.curproc.rlimit.stacksize.hard = 67108864
proc.curproc.rlimit.coredumpsize.soft = unlimited
proc.curproc.rlimit.coredumpsize.hard = unlimited
proc.curproc.rlimit.memoryuse.soft = 2094112768
proc.curproc.rlimit.memoryuse.hard = 2094112768
proc.curproc.rlimit.memorylocked.soft = 698037589
proc.curproc.rlimit.memorylocked.hard = 2094112768
proc.curproc.rlimit.maxproc.soft = 160
(Continue reading)

Permalink | Reply | 1 comment |
headers
Izumi Tsutsui | 18 Jun 13:19
Picon

Re: port-i386/43156: NetBSD bootloader countdown runs at 1/20 speed in qemu 0.12

Hi,

Is there any objection to commiting this bootloader changes for QEMU
in PR port-i386/43156?
---
Izumi Tsutsui

--- original message ---

> >Synopsis:       NetBSD bootloader countdown runs at 1/20 speed in qemu 0.12

> After upgrading the qemu package to version 0.12, the bootloader
> countdown in a qemu emulated NetBSD system takes about 20 times longer
> than it should, counting down by one every 20 seconds instead of every
> second, taking a total of 100 seconds to count down from five.

Probably BIOS in the new QEMU distribution implements a BIOS function call
INT 15h/AH=86h (WAIT), but the emulation takes more time than it should
on long delays.

Fixing BIOS package in QEMU is not so easy so it would be better
to apply workaround using INT 1Ah/AH=00h (GET SYSTEMTIME)
as used to fix the similar countdown speed issue:
http://mail-index.NetBSD.org/source-changes/2009/08/26/msg000190.html

The attached diff seems working for me but I wonder if we should have
more common API to wait specified seconds in libsa...

Index: lib/biosdisk.c
===================================================================
(Continue reading)

Permalink | Reply | 0 comments |
headers
Jean-Yves Migeon | 10 Jun 01:43
Picon
Favicon

[PAE support] Initial patch review

Dear all,

Here is a patch [1] that "ports" PAE support from Xen to GENERIC.

Currently, the patch triggers a double fault some time after boot with 
dom0 PAE under Xen. As such, it is not yet ready for commit, but I 
consider it mature enough to ask for an initial review.

It diverges quite a bit from the original patch from Jeremy. Reason is 
to avoid too many #ifdef's between Xen and non-Xen pmap (all the 
delightful details are in the comments inside the patch - in short, Xen 
tracks reference counts to L3 kernel page + we cannot use recursive 
mappings easily). It consumes 4kB for each CPU attached, instead of each 
process created.

Here is a summary of what the patch does; any advice on it will be 
appreciated. I tested it under QEMU; unfortunately, I can't stress test 
MP code, I am not confident on the -smp flag's capability on a mono-core 
host.

- principle: the PTP_LEVELS remains at 2. The L3 page is a page 
allocated per-CPU (below the 4GB boundary, due to %cr3 size limitation), 
and the kernel keeps track of the allocation through 2 additional struct 
cpu_info elements:
   - ci_l3_pdir for the virtual address of the PD,
   - ci_l3_pdirpa for its PA counterpart

- context switch with PAE is a matter of editing the 4 L3 entries, not 
changing %cr3 value (this is the non-PAE situation).
(Continue reading)

Permalink | Reply | 2 comments |
headers
Prudential Finance VietNam Co.Ltd | 8 Jun 23:01

VAY TIEU DUNG - THU TUC NHANH GON


                             CHUONG TRINH HO TRO: VAY TIN CHAP - TIEU DUNG CA NHAN

    Nham dap ung nhu cau ve tai chinh cua nhieu can bo cong nhan vien chuc thuoc cac to chuc, doanh nghiep tren dia
ban Ha Noi va TP Ho Chi Minh, Cty TNHH  Prudential Finance VietNam da cho ra doi va cung cap dich vu  TIN DUNG TIEU
DUNG voi hinh thuc va thoi gian linh hoat, lai suat hap dan.

 1. Uu diem
-	Lai suat uu dai, tu van mien phi
-	Khong can the chap, tai san dam bao
-	Khong can bao lanh cua co quan. Lanh dao doanh nghiep
-	Thu tuc don gian, khong can cong chung giay to ho so (co the gui ho so qua Fax)
	
2. Dieu kien
-	Ca nhan co ho khau hoac KT3/KT4 tai TP Ho Chi Minh va Ha Noi (bao gom ca Ha Tay cu)
-	Hien dang lam viec tai cac don vi, to chuc, doanh nghiep nha nuoc, cong ty co phan, cong ty nuoc ngoai, lien
doanh, van phong dai dien tai TP Ho Chi Minh va Ha Noi co muc thu nhap toi thieu 3 Trieu VND
-	Khach hang kinh doanh ( Co giay phep dang ky kinh doanh tu 2 nam tro len)
	
3. Han muc, lai suat vay
-	Tu 10.000.000 den 190.000.000 VN?
-	Lai vay tuy thuoc dieu kien khach hang
-	Lai suat:14.9%/Nam ' 22.1%/Nam
	
4. Thoi gian vay
-	Tu 12 thang den 48 thang
	
5. Thoi gian giai ngan
-	03- 05 ngay sau khi hoan tat ho so
(Continue reading)

Permalink | Reply | 0 comments |
headers
Steve Blinkhorn | 8 Jun 13:36
Picon
Favicon

onboard modem support?

Is there any support for SIS product 0x7013, onboard modem, or is this
one of those Windoze-only stripped-down devices?

--

-- 
Steve Blinkhorn <steve <at> prd.co.uk>
Permalink | Reply | 0 comments |
headers
Teri debugger | 4 Jun 10:24
Favicon

business list

Here's some of the 30 lists I have Doctors-Healthcare-Business–Finance Professionals & Consumers:

Criminal Attorneys - 142,906 total records, 99,857 emails
USA Lawyers Database - 269,787 records with 235,244 emails
Financial Planners Database - 148,857 records all with emails	
Finance and Money Professionals Database - 116,568 records all with emails

All complete lists above: $299

There are more list packages also available for consumer, healthcare, business and more. Contact me here
for more info or to get samples: Jordan.Stephenson <at> listparadise.info

to terminate please send a blank message to exit <at> listparadise.info
Permalink | Reply | 0 comments |

Gmane