2002-09-18 13:01:55

by Olaf Frączyk

[permalink] [raw]
Subject: Which processor/board for embedded NTP

Hi All,

I have to build NTP server on an embedded pc.
Which processors/boards are suitable for this.
Such processor/board cannot have ANY problems with time handling.
I thought about Geode (NS), but I found some info that it doesn't have
TSC, or it works not properly (the PPS needs TSC).
And of course has to work excellent with linux.

Please CC me, as I'm not subscribed.

Best Regards,

Olaf Fraczyk






2002-09-18 13:50:16

by Alan

[permalink] [raw]
Subject: Re: Which processor/board for embedded NTP

On Wed, 2002-09-18 at 14:10, Olaf Fr?czyk wrote:
> I have to build NTP server on an embedded pc.
> Which processors/boards are suitable for this.
> Such processor/board cannot have ANY problems with time handling.

Game over. There is no PC technology that doesn't have some problems
with time handling 8)

> I thought about Geode (NS), but I found some info that it doesn't have
> TSC, or it works not properly (the PPS needs TSC).
> And of course has to work excellent with linux.

The Geode aka Cyrix MediaGx TSC problems are not in the TSC but in the
PIT emulation in the chipset. That problem isnt present on the 5530 and
later chipsets. So its a long buried problem. You might also want to
look at the VIA Eden based boards if its got to be embedded (fanless
550MHz)

2002-09-18 14:19:49

by Kirk Reiser

[permalink] [raw]
Subject: Athlon/via problems continued

Hi Folks: Here are a couple portions of boot messages and a bug()
trigger from 2.5.36. I am not sure what is important so forgive me
for extraneous info.

Sep 18 08:40:10 itchy kernel: Linux version 2.5.36 (kirk@itchy) (gcc version 2.95.4 20011002 (Debian prerelease)) #0 Wed Sep 18 08:28:42 EDT 2002
Sep 18 08:40:10 itchy kernel: Video mode to be used for restore is f01
Sep 18 08:40:10 itchy kernel: BIOS-provided physical RAM map:
Sep 18 08:40:10 itchy kernel: BIOS-e820: 0000000000000000 - 00000000000a0000 (usable)
Sep 18 08:40:10 itchy kernel: BIOS-e820: 00000000000f0000 - 0000000000100000 (reserved)
Sep 18 08:40:10 itchy kernel: BIOS-e820: 0000000000100000 - 0000000020000000 (usable)
Sep 18 08:40:10 itchy kernel: BIOS-e820: 00000000ffff0000 - 0000000100000000 (reserved)
Sep 18 08:40:10 itchy kernel: 512MB LOWMEM available.
Sep 18 08:40:10 itchy kernel: On node 0 totalpages: 131072
Sep 18 08:40:10 itchy kernel: DMA zone: 4096 pages
Sep 18 08:40:10 itchy kernel: Normal zone: 126976 pages
Sep 18 08:40:10 itchy kernel: HighMem zone: 0 pages
Sep 18 08:40:10 itchy kernel: Kernel command line: BOOT_IMAGE=old ro root=301
Sep 18 08:40:10 itchy kernel: Initializing CPU#0
Sep 18 08:40:10 itchy kernel: Detected 1737.233 MHz processor.
Sep 18 08:40:10 itchy kernel: Console: colour VGA+ 80x50
Sep 18 08:40:10 itchy kernel: Calibrating delay loop... 3424.25 BogoMIPS
Sep 18 08:40:10 itchy kernel: Memory: 516800k/524288k available (1097k kernel code, 7104k reserved, 404k data, 224k init, 0k highmem)
Sep 18 08:40:10 itchy kernel: Security Scaffold v1.0.0 initialized
Sep 18 08:40:10 itchy kernel: Dentry-cache hash table entries: 65536 (order: 7, 524288 bytes)
Sep 18 08:40:10 itchy kernel: Inode-cache hash table entries: 32768 (order: 6, 262144 bytes)
Sep 18 08:40:10 itchy kernel: Mount-cache hash table entries: 512 (order: 0, 4096 bytes)
Sep 18 08:40:10 itchy kernel: CPU: Before vendor init, caps: 0383f9ff c1c3f9ff 00000000, vendor = 2
Sep 18 08:40:10 itchy kernel: CPU: L1 I Cache: 64K (64 bytes/line), D cache 64K (64 bytes/line)
Sep 18 08:40:10 itchy kernel: CPU: L2 Cache: 256K (64 bytes/line)
Sep 18 08:40:10 itchy kernel: CPU: After vendor init, caps: 0383f9ff c1c3f9ff 00000000 00000000
Sep 18 08:40:10 itchy kernel: Intel machine check architecture supported.
Sep 18 08:40:10 itchy kernel: Intel machine check reporting enabled on CPU#0.
Sep 18 08:40:10 itchy kernel: Machine check exception polling timer started.
Sep 18 08:40:10 itchy kernel: CPU: After generic, caps: 0383f9ff c1c3f9ff 00000000 00000000
Sep 18 08:40:10 itchy kernel: CPU: Common caps: 0383f9ff c1c3f9ff 00000000 00000000
Sep 18 08:40:10 itchy kernel: CPU: AMD Athlon(tm) XP 2100+ stepping 02
Sep 18 08:40:10 itchy kernel: Enabling fast FPU save and restore... done.
Sep 18 08:40:10 itchy kernel: Enabling unmasked SIMD FPU exception support... done.
Sep 18 08:40:10 itchy kernel: Checking 'hlt' instruction... OK.
Sep 18 08:40:10 itchy kernel: POSIX conformance testing by UNIFIX
Sep 18 08:40:10 itchy kernel: Linux NET4.0 for Linux 2.4
Sep 18 08:40:10 itchy kernel: Based upon Swansea University Computer Society NET3.039
Sep 18 08:40:10 itchy kernel: Initializing RT netlink socket
Sep 18 08:40:10 itchy kernel: PCI: PCI BIOS revision 2.10 entry at 0xfb490, last bus=1
Sep 18 08:40:10 itchy kernel: PCI: Using configuration type 1
Sep 18 08:40:10 itchy kernel: PCI: Probing PCI hardware
Sep 18 08:40:10 itchy kernel: PCI: Probing PCI hardware (bus 00)
Sep 18 08:40:10 itchy kernel: PCI: Using IRQ router default [1106/3099] at 00:00.0
Sep 18 08:40:10 itchy kernel: Starting kswapd
Sep 18 08:40:10 itchy kernel: BIO: pool of 256 setup, 14Kb (56 bytes/bio)
Sep 18 08:40:10 itchy kernel: biovec pool[0]: 1 bvecs: 256 entries (12 bytes)
Sep 18 08:40:10 itchy kernel: biovec pool[1]: 4 bvecs: 256 entries (48 bytes)
Sep 18 08:40:10 itchy kernel: biovec pool[2]: 16 bvecs: 256 entries (192 bytes)
Sep 18 08:40:10 itchy kernel: biovec pool[3]: 64 bvecs: 256 entries (768 bytes)
Sep 18 08:40:10 itchy kernel: biovec pool[4]: 128 bvecs: 256 entries (1536 bytes)
Sep 18 08:40:10 itchy kernel: biovec pool[5]: 256 bvecs: 256 entries (3072 bytes)
Sep 18 08:40:10 itchy kernel: aio_setup: sizeof(struct page) = 40
Sep 18 08:40:10 itchy kernel: Journalled Block Device driver loaded
Sep 18 08:40:10 itchy kernel: Capability LSM initialized
Sep 18 08:40:10 itchy kernel: block: 256 slots per queue, batch=32
Sep 18 08:40:10 itchy kernel: Floppy drive(s): fd0 is 1.44M
Sep 18 08:40:10 itchy kernel: FDC 0 is a post-1991 82077
Sep 18 08:40:10 itchy kernel: 3c59x: Donald Becker and others. http://www.scyld.com/network/vortex.html
Sep 18 08:40:10 itchy kernel: 00:09.0: 3Com PCI 3c905C Tornado at 0xd000. Vers LK1.1.18
Sep 18 08:40:10 itchy kernel: phy=0, phyx=24, mii_status=0x782d
Sep 18 08:40:10 itchy kernel: Uniform Multi-Platform E-IDE driver Revision: 7.00alpha2
Sep 18 08:40:10 itchy kernel: ide: Assuming 33MHz system bus speed for PIO modes; override with idebus=xx
Sep 18 08:40:10 itchy kernel: VP_IDE: IDE controller at PCI slot 00:11.1
Sep 18 08:40:10 itchy kernel: PCI: No IRQ known for interrupt pin A of device 00:11.1. Please try using pci=biosirq.
Sep 18 08:40:10 itchy kernel: VP_IDE: chipset revision 6
Sep 18 08:40:10 itchy kernel: VP_IDE: not 100%% native mode: will probe irqs later
Sep 18 08:40:10 itchy kernel: ide: Assuming 33MHz system bus speed for PIO modes; override with idebus=xx
Sep 18 08:40:10 itchy kernel: VP_IDE: VIA vt8233a (rev 00) IDE UDMA133 controller on pci00:11.1
Sep 18 08:40:10 itchy kernel: ide0: BM-DMA at 0xd400-0xd407, BIOS settings: hda:DMA, hdb:pio
Sep 18 08:40:10 itchy kernel: ide1: BM-DMA at 0xd408-0xd40f, BIOS settings: hdc:DMA, hdd:pio
Sep 18 08:40:10 itchy kernel: hda: FUJITSU MPG3409AT E, ATA DISK drive
Sep 18 08:40:10 itchy kernel: hda: DMA disabled
Sep 18 08:40:10 itchy kernel: ide0 at 0x1f0-0x1f7,0x3f6 on irq 14
Sep 18 08:40:10 itchy kernel: hdc: HL-DT-ST GCE-8320B, ATAPI CD/DVD-ROM drive
Sep 18 08:40:10 itchy kernel: hdc: DMA disabled
Sep 18 08:40:10 itchy kernel: ide1 at 0x170-0x177,0x376 on irq 15
Sep 18 08:40:10 itchy kernel: VP_IDE: IDE controller at PCI slot 00:11.1
Sep 18 08:40:10 itchy kernel: PCI: No IRQ known for interrupt pin A of device 00:11.1. Please try using pci=biosirq.
Sep 18 08:40:10 itchy kernel: PCI: Unable to reserve I/O region #5:10@d400 for device 00:11.1
Sep 18 08:40:10 itchy kernel: VP_IDE: not 100%% native mode: will probe irqs later
Sep 18 08:40:10 itchy kernel: VP_IDE: VIA vt8233a (rev 00) IDE UDMA133 controller on pci00:11.1
Sep 18 08:40:10 itchy kernel: VP_IDE: port 0x01f0 already claimed by ide0
Sep 18 08:40:10 itchy kernel: VP_IDE: port 0x0170 already claimed by ide1
Sep 18 08:40:10 itchy kernel: VP_IDE: neither IDE port enabled (BIOS)
Sep 18 08:40:10 itchy kernel: hda: host protected area => 1
Sep 18 08:40:10 itchy kernel: hda: 80063424 sectors (40992 MB) w/2048KiB Cache, CHS=4983/255/63, UDMA(100)
Sep 18 08:40:10 itchy kernel: hda: hda1 hda2 hda3 hda4
Sep 18 08:40:10 itchy kernel: hdc: ATAPI 40X CD-ROM CD-R/RW drive, 8192kB Cache, DMA
Sep 18 08:40:10 itchy kernel: Uniform CD-ROM driver Revision: 3.12
Sep 18 08:40:10 itchy kernel: serio: i8042 AUX port at 0x60,0x64 irq 12
Sep 18 08:40:10 itchy kernel: input: AT Set 2 keyboard on isa0060/serio0
Sep 18 08:40:10 itchy kernel: serio: i8042 KBD port at 0x60,0x64 irq 1
Sep 18 08:40:10 itchy kernel: NET4: Linux TCP/IP 1.0 for NET4.0
Sep 18 08:40:10 itchy kernel: IP Protocols: ICMP, UDP, TCP, IGMP
Sep 18 08:40:10 itchy kernel: IP: routing cache hash table of 4096 buckets, 32Kbytes
Sep 18 08:40:10 itchy kernel: TCP: Hash tables configured (established 131072 bind 65536)
Sep 18 08:40:10 itchy kernel: NET4: Unix domain sockets 1.0/SMP for Linux NET4.0.
Sep 18 08:40:10 itchy kernel: kjournald starting. Commit interval 5 seconds
Sep 18 08:40:10 itchy kernel: EXT3-fs: mounted filesystem with ordered data mode.
Sep 18 08:40:10 itchy kernel: VFS: Mounted root (ext3 filesystem) readonly.
Sep 18 08:40:10 itchy kernel: Freeing unused kernel memory: 224k freed
Sep 18 08:40:10 itchy kernel: Adding 136544k swap on /dev/hda2. Priority:-1 extents:1
Sep 18 08:40:10 itchy kernel: EXT3 FS 2.4-0.9.16, 02 Dec 2001 on ide0(3,1), internal journal
Sep 18 08:40:10 itchy kernel: kjournald starting. Commit interval 5 seconds
Sep 18 08:40:10 itchy kernel: EXT3 FS 2.4-0.9.16, 02 Dec 2001 on ide0(3,3), internal journal
Sep 18 08:40:10 itchy kernel: EXT3-fs: mounted filesystem with ordered data mode.
Sep 18 08:40:10 itchy kernel: kjournald starting. Commit interval 5 seconds
Sep 18 08:40:10 itchy kernel: EXT3 FS 2.4-0.9.16, 02 Dec 2001 on ide0(3,4), internal journal
Sep 18 08:40:10 itchy kernel: EXT3-fs: mounted filesystem with ordered data mode.

Here is the pci/ide portion of the next run with pci=boisirq specified
on the command line.

Sep 18 09:24:24 itchy kernel: PCI: PCI BIOS revision 2.10 entry at 0xfb490, last bus=1
Sep 18 09:24:24 itchy kernel: PCI: Using configuration type 1
Sep 18 09:24:24 itchy kernel: PCI: Probing PCI hardware
Sep 18 09:24:24 itchy kernel: PCI: Probing PCI hardware (bus 00)
Sep 18 09:24:24 itchy kernel: PCI: Using IRQ router default [1106/3099] at 00:00.0
Sep 18 09:24:24 itchy kernel: ide: Assuming 33MHz system bus speed for PIO modes; override with idebus=xx
Sep 18 09:24:24 itchy kernel: VP_IDE: IDE controller at PCI slot 00:11.1
Sep 18 09:24:24 itchy kernel: PCI: No IRQ known for interrupt pin A of device 00:11.1.
Sep 18 09:24:24 itchy kernel: VP_IDE: chipset revision 6
Sep 18 09:24:24 itchy kernel: VP_IDE: not 100%% native mode: will probe irqs later
Sep 18 09:24:24 itchy kernel: ide: Assuming 33MHz system bus speed for PIO modes; override with idebus=xx
Sep 18 09:24:24 itchy kernel: VP_IDE: VIA vt8233a (rev 00) IDE UDMA133 controller on pci00:11.1
Sep 18 09:24:24 itchy kernel: ide0: BM-DMA at 0xd400-0xd407, BIOS settings: hda:DMA, hdb:pio
Sep 18 09:24:24 itchy kernel: ide1: BM-DMA at 0xd408-0xd40f, BIOS settings: hdc:DMA, hdd:pio
Sep 18 09:24:24 itchy kernel: hda: FUJITSU MPG3409AT E, ATA DISK drive
Sep 18 09:24:24 itchy kernel: hda: DMA disabled
Sep 18 09:24:24 itchy kernel: ide0 at 0x1f0-0x1f7,0x3f6 on irq 14
Sep 18 09:24:24 itchy kernel: hdc: HL-DT-ST GCE-8320B, ATAPI CD/DVD-ROM drive
Sep 18 09:24:24 itchy kernel: hdc: DMA disabled
Sep 18 09:24:24 itchy kernel: ide1 at 0x170-0x177,0x376 on irq 15
Sep 18 09:24:24 itchy kernel: VP_IDE: IDE controller at PCI slot 00:11.1
Sep 18 09:24:24 itchy kernel: PCI: No IRQ known for interrupt pin A of device 00:11.1.
Sep 18 09:24:24 itchy kernel: PCI: Unable to reserve I/O region #5:10@d400 for device 00:11.1
Sep 18 09:24:24 itchy kernel: VP_IDE: not 100%% native mode: will probe irqs later
Sep 18 09:24:24 itchy kernel: VP_IDE: VIA vt8233a (rev 00) IDE UDMA133 controller on pci00:11.1
Sep 18 09:24:24 itchy kernel: VP_IDE: port 0x01f0 already claimed by ide0
Sep 18 09:24:24 itchy kernel: VP_IDE: port 0x0170 already claimed by ide1
Sep 18 09:24:24 itchy kernel: VP_IDE: neither IDE port enabled (BIOS)
Sep 18 09:24:24 itchy kernel: hda: host protected area => 1
Sep 18 09:24:24 itchy kernel: hda: 80063424 sectors (40992 MB) w/2048KiB Cache, CHS=4983/255/63, UDMA(100)
Sep 18 09:24:24 itchy kernel: hda: hda1 hda2 hda3 hda4
Sep 18 09:24:24 itchy kernel: hdc: ATAPI 40X CD-ROM CD-R/RW drive, 8192kB Cache, DMA
Sep 18 09:24:24 itchy kernel: Uniform CD-ROM driver Revision: 3.12
Sep 18 09:24:24 itchy kernel: serio: i8042 AUX port at 0x60,0x64 irq 12
I don't understand why the vpide would allow the standard ide to keep
the ports 0x1f0 if the vpide could handle the controller better for
the via chip set. I also don't understand why the DMA was turned
off. Interestingly when I ran hdparm -vi it showed DMA as being
enabled.

Now, I run a chron job every twenty minutes to attempt to start
setiathome incase it has been stopped for some reason. This is
usually the point where the box takes a trip south, as in solid lock
up. I managed to get a bug() dump for once. If I need to supply more
info please let me know what else is necessary.

Sep 18 09:40:59 itchy kernel: kernel BUG at page_alloc.c:184!
Sep 18 09:40:59 itchy kernel: invalid operand: 0000
Sep 18 09:40:59 itchy kernel: CPU: 0
Sep 18 09:40:59 itchy kernel: EIP: 0060:[rmqueue+646/784] Not tainted
Sep 18 09:40:59 itchy kernel: EFLAGS: 00010202
Sep 18 09:40:59 itchy kernel: eax: 01010024 ebx: c14d5228 ecx: 00000000 edx: 01010024
Sep 18 09:40:59 itchy kernel: esi: c0269b60 edi: c0269b60 ebp: 00000001 esp: df14be40
Sep 18 09:40:59 itchy kernel: ds: 0068 es: 0068 ss: 0068
Sep 18 09:40:59 itchy kernel: Process setiathome (pid: 215, threadinfo=df14a000 task=df37e200)
Sep 18 09:40:59 itchy kernel: Stack: 000001ff 00000000 c0269d40 410a5000 00000000 c0107160 00001000 0001deda
Sep 18 09:40:59 itchy kernel: 00000286 c0269b98 c0129d82 c0269b60 00000000 000001d2 c10028b8 deb8c294
Sep 18 09:40:59 itchy kernel: 410a5000 c0269b60 00000000 000001d2 1eedb067 c0129b46 00104025 c01208e6
Sep 18 09:40:59 itchy kernel: Call Trace: [common_interrupt+24/32] [__alloc_pages+82/480] [_alloc_pages+22/32] [do_anonymous_page+70/352] [do_no_page+52/528]
Sep 18 09:40:59 itchy kernel: [handle_mm_fault+97/224] [do_page_fault+279/1028] [do_page_fault+0/1028] [update_process_times+37/48] [update_wall_time+11/64] [timer_bh+36/624]
Sep 18 09:40:59 itchy kernel: [bh_action+19/32] [tasklet_hi_action+61/112] [do_softirq+90/176] [do_IRQ+200/224] [error_code+45/56]
Sep 18 09:40:59 itchy kernel:
Sep 18 09:40:59 itchy kernel: Code: 0f 0b b8 00 c0 bd 21 c0 89 f6 8b 03 a8 40 74 0a 0f 0b b9 00

Kirk

--

Kirk Reiser The Computer Braille Facility
e-mail: [email protected] University of Western Ontario
phone: (519) 661-3061

2002-09-18 19:32:03

by Christer Weinigel

[permalink] [raw]
Subject: Re: Which processor/board for embedded NTP

Olaf Fr?czyk <[email protected]> writes:

> I have to build NTP server on an embedded pc.
> Which processors/boards are suitable for this.
> Such processor/board cannot have ANY problems with time handling.
> I thought about Geode (NS), but I found some info that it doesn't have
> TSC, or it works not properly (the PPS needs TSC).
> And of course has to work excellent with linux.

The Geode's TSC will work quite well for you if you disable the
suspend on HLT option in the processor.

On the newer "IA on a chip" geodes (SC1200, SC2200 and SC3200) there
is also a high speed timer in the chipset that seems to be quite
stable.

/Christer

--
"Just how much can I get away with and still go to heaven?"

Freelance consultant specializing in device driver programming for Linux
Christer Weinigel <[email protected]> http://www.weinigel.se

2002-09-19 05:57:18

by Robert Schwebel

[permalink] [raw]
Subject: Re: Which processor/board for embedded NTP

Hi Christer,

On Wed, Sep 18, 2002 at 09:37:02PM +0200, Christer Weinigel wrote:
> On the newer "IA on a chip" geodes (SC1200, SC2200 and SC3200) there
> is also a high speed timer in the chipset that seems to be quite
> stable.

Did the realtime behaviour of the newer Geodes also change to the
better? Last time we tested them with RTAI there have been Buslocks
which prevented them from being usefull...

Robert
--
Dipl.-Ing. Robert Schwebel | http://www.pengutronix.de
Pengutronix - Linux Solutions for Science and Industry
Braunschweiger Str. 79, 31134 Hildesheim, Germany
Handelsregister: Amtsgericht Hildesheim, HRA 2686
Phone: +49-5121-28619-0 | Fax: +49-5121-28619-4

2002-09-19 19:34:11

by Christer Weinigel

[permalink] [raw]
Subject: MediaGX/Geode performance fix, Was: Which processor/board for embedded NTP

This mail contains a patch to fix a performance problem with many
Cyrix MediaGX/NatSemi Geode platforms. The register settings have
been officially recommended by NatSemi themselves. The patch is
against linux-2.4.20-pre7. Should this be merged into the mainsteam
linux kernel?

Robert Schwebel <[email protected]> writes:

> Hi Christer,
>
> On Wed, Sep 18, 2002 at 09:37:02PM +0200, Christer Weinigel wrote:
> > On the newer "IA on a chip" geodes (SC1200, SC2200 and SC3200) there
> > is also a high speed timer in the chipset that seems to be quite
> > stable.
>
> Did the realtime behaviour of the newer Geodes also change to the
> better? Last time we tested them with RTAI there have been Buslocks
> which prevented them from being usefull...

Speaking about this, there is a bug in many BIOS:es for the Cyrix
MediaGX and Natsemi Geode that according to the BIOS errata will lower
the performance of master PCI transfers from 70MB/s to 25MB/s.

What happens is that the bit SDBE is set to 1 when it should be set to
0, causing PCI disconnects every 16 bytes which is slow.

Index 41h PCI Control Function 2 Register (R/W)

3:2 SDB Slave Disconnect Boundary: PCI slave issues a disconnect with
data when it crosses line boundary:
00 = 128 bytes
01 = 256 bytes
10 = 512 bytes
11 = 1024 bytes
Works in conjunction with bit 1.

1 SDBE Slave Disconnect Boundary Enable:
0 = PCI slave disconnects on boundaries set by bits [3:2].
1 = PCI disconnects on cache line boundary which is 16 bytes.

So, this patch just sets the SDBE bit to 0 and prints a message to
that effect.

/Christer

diff -ur linux/drivers/pci/quirks.c.orig linux/drivers/pci/quirks.c
--- linux/drivers/pci/quirks.c.orig Thu Sep 19 21:34:21 2002
+++ linux/drivers/pci/quirks.c Thu Sep 19 21:35:07 2002
@@ -477,6 +477,24 @@
}

/*
+ * Common misconfiguration of the MediaGX/Geode PCI master that will
+ * reduce PCI bandwidth from 70MB/s to 25MB/s. See the GXM/GXLV/GX1
+ * datasheets found at http://www.national.com/ds/GX for info on what
+ * these bits do. <[email protected]>
+ */
+
+static void __init quirk_mediagx_master(struct pci_dev *dev)
+{
+ u8 reg;
+ pci_read_config_byte(dev, 0x41, &reg);
+ if (reg & 2) {
+ reg &= ~2;
+ printk(KERN_INFO "PCI: Fixup for MediaGX/Geode Slave Disconnect Boundary (0x41=0x%02x)\n", reg);
+ pci_write_config_byte(dev, 0x41, reg);
+ }
+}
+
+/*
* The main table of quirks.
*/

@@ -538,6 +556,8 @@
*/
{ PCI_FIXUP_HEADER, PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_82380FB, quirk_transparent_bridge },

+ { PCI_FIXUP_FINAL, PCI_VENDOR_ID_CYRIX, PCI_DEVICE_ID_CYRIX_PCI_MASTER, quirk_mediagx_master },
+
{ 0 }
};

--
"Just how much can I get away with and still go to heaven?"

Freelance consultant specializing in device driver programming for Linux
Christer Weinigel <[email protected]> http://www.weinigel.se

2002-09-19 23:02:36

by Alan

[permalink] [raw]
Subject: Re: MediaGX/Geode performance fix, Was: Which processor/board for embedded NTP

On Thu, 2002-09-19 at 20:39, Christer Weinigel wrote:
> This mail contains a patch to fix a performance problem with many
> Cyrix MediaGX/NatSemi Geode platforms. The register settings have
> been officially recommended by NatSemi themselves. The patch is
> against linux-2.4.20-pre7. Should this be merged into the mainsteam
> linux kernel?

This wont actually make an iota of difference in most cases. The CS5530
IDE driver will force this value to 0x14 anyway. It also sets MWI on te
X-bus which is needed too.

Probably the fixup should be done in the PCI quirks.

2002-09-20 09:54:39

by Christer Weinigel

[permalink] [raw]
Subject: Re: MediaGX/Geode performance fix, Was: Which processor/board for embedded NTP

Alan Cox <[email protected]> writes:

> On Thu, 2002-09-19 at 20:39, Christer Weinigel wrote:
> > This mail contains a patch to fix a performance problem with many
> > Cyrix MediaGX/NatSemi Geode platforms. The register settings have
> > been officially recommended by NatSemi themselves. The patch is
> > against linux-2.4.20-pre7. Should this be merged into the mainsteam
> > linux kernel?
>
> This wont actually make an iota of difference in most cases. The CS5530
> IDE driver will force this value to 0x14 anyway. It also sets MWI on te
> X-bus which is needed too.

Ok, it did make a 5% difference when I did some quick tests on TCP
performance. But I didn't have the CS5530 IDE driver loaded.

> Probably the fixup should be done in the PCI quirks.

Sounds it should be moved from the IDE driver to quirks at least, but
that might be a 2.5 thing in that case.

/Christer

--
"Just how much can I get away with and still go to heaven?"

Freelance consultant specializing in device driver programming for Linux
Christer Weinigel <[email protected]> http://www.weinigel.se