Re: Status of ALi MAGiK 1 support in 2.4.?
On Tuesday 29 May 2001 11:04, Bobby D. Bryant wrote: > Alan Cox wrote: > > > May 22 21:45:07 pollux kernel: ALI15X3: simplex device: DMA disabled > > > May 22 21:45:07 pollux kernel: ide0: ALI15X3 Bus-Master DMA disabled > > > (BIOS) > > > May 22 21:45:07 pollux kernel: ALI15X3: simplex device: DMA disabled > > > May 22 21:45:07 pollux kernel: ide1: ALI15X3 Bus-Master DMA disabled > > > > The DMA was off because the BIOS left it off. > > I just checked, and the BIOS auto-detect page for that drive shows PIO Mode > 4 and Ultra DMA Mode 5. The BIOS also shows a summary chart during boot, > just before the LILO prompt, and that summary also reports UDMA 5 for that > drive. It really looks like the kernel is not getting the correct device > info from the BIOS. Just a blind guess, but this reminds me of a similar problem I had with a different Athlon board a while ago. The BIOS and the driver disagreed on the bits through which DMA capability was signalled. The fix that worked for me was to enable both IDE channels (and use up two interrupts) though I needed only the first channel. -- Stefan R. Jaschke <[EMAIL PROTECTED]> http://www.jaschke-net.de - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: Status of ALi MAGiK 1 support in 2.4.?
Mike Frisch wrote: > On Tue, May 29, 2001 at 05:18:39AM +0600, Bobby D. Bryant wrote: > > sailing ever since. The only problems that I'm are ware of are a > > (maybe) DMA problem and a (maybe) SMBus problem, per below. Right now > > I noticed the Win32 benchmark/test application Sandra mentioned an SMBus > problem with the A7A266 as well. I have yet to try lm_sensors myself, > but it looks like I won't get far. My sensors are working and appear to give the correct numbers; I'm just getting a regular spew of messages from the SMB resets. > > May 22 21:45:07 pollux kernel: ALI15X3: IDE controller on PCI bus 00 dev > > 20 > > May 22 21:45:07 pollux kernel: PCI: No IRQ known for interrupt pin A of > > device 00:04.0. Please try using pci=biosirq. > > I get the same message, but it does not appear to dramatically affect > my performance. As I mentioned, I am getting 25MB/s (through hdparm; I > have yet to try anything more) with my Quantum Fireball. My DMA is > enabled in the BIOS and detected by the kernel. That's interesting, in light of the other message I just sent (i.e., DMA enabled in BIOS, but *not* detected by the kernel). What version of the BIOS are you running? I'm still on the 1003b, because I've heard a few horror stories about 1004 that have put me off upgrading. I'm on kernel 2.4.4, with Athlon optimizations. > > The routing to IQR 0 sounds funny to me, but this is already way beyond > > what I understand. > > Do you have the PnP operating system setting in the BIOS turned off? > (ie. telling the BIOS you have non-PnP aware O/S) I noticed that prior > to doing this, all of my PCI cards were listed as IRo 0. I think I tried it both ways earlier. At any rate, since I had to reboot to check the DMA settings I went ahead and set it back to 'off', and it still says the same thing. Thanks, Bobby Bryant Austin, Texas - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: Status of ALi MAGiK 1 support in 2.4.?
Alan Cox wrote: > > May 22 21:45:07 pollux kernel: ALI15X3: simplex device: DMA disabled > > May 22 21:45:07 pollux kernel: ide0: ALI15X3 Bus-Master DMA disabled > > (BIOS) > > May 22 21:45:07 pollux kernel: ALI15X3: simplex device: DMA disabled > > May 22 21:45:07 pollux kernel: ide1: ALI15X3 Bus-Master DMA disabled > > The DMA was off because the BIOS left it off. I just checked, and the BIOS auto-detect page for that drive shows PIO Mode 4 and Ultra DMA Mode 5. The BIOS also shows a summary chart during boot, just before the LILO prompt, and that summary also reports UDMA 5 for that drive. It really looks like the kernel is not getting the correct device info from the BIOS. As I mentioned earlier, the A7A266 supposedly has an ALi M1535D+ southbridge "with PCI Super-I/O Integrated Peripheral Controller (PSIPC)", rather than the ALI15X3 reported by the kernel. Thanks, Bobby Bryant Austin, Texas - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: Status of ALi MAGiK 1 support in 2.4.?
On Tue, May 29, 2001 at 05:18:39AM +0600, Bobby D. Bryant wrote: > sailing ever since. The only problems that I'm are ware of are a > (maybe) DMA problem and a (maybe) SMBus problem, per below. Right now I noticed the Win32 benchmark/test application Sandra mentioned an SMBus problem with the A7A266 as well. I have yet to try lm_sensors myself, but it looks like I won't get far. > May 22 21:45:07 pollux kernel: ALI15X3: IDE controller on PCI bus 00 dev > 20 > May 22 21:45:07 pollux kernel: PCI: No IRQ known for interrupt pin A of > device 00:04.0. Please try using pci=biosirq. I get the same message, but it does not appear to dramatically affect my performance. As I mentioned, I am getting 25MB/s (through hdparm; I have yet to try anything more) with my Quantum Fireball. My DMA is enabled in the BIOS and detected by the kernel. > The routing to IQR 0 sounds funny to me, but this is already way beyond > what I understand. Do you have the PnP operating system setting in the BIOS turned off? (ie. telling the BIOS you have non-PnP aware O/S) I noticed that prior to doing this, all of my PCI cards were listed as IRo 0. Mike. - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: Status of ALi MAGiK 1 support in 2.4.?
Mike Frisch wrote: > On Mon, May 28, 2001 at 05:57:12PM +0200, Axel Thimm wrote: > > What is the status of the support for this chipset, found for example in an > > ASUS A7A266? Judging from > > http://www.acerlabs.com/eng/support/faqlnx.htm > > one gets the impression that ALi is respectfully treating the Linux community. > > I cannot answer your question about the level of support this chipset > has, but suffice it to say that my new (as of last week) A7A266 based > system (1.2Ghz T-Bird w/256MB Crucial DDR RAM) is running 2.4.5 (and > previous to that 2.2.18) quite nicely. > I finally jerked the braindead 8KTA3 out of one of my new systems and replaced it with an A7A266 about two weeks ago, and it has been smooth sailing ever since. The only problems that I'm are ware of are a (maybe) DMA problem and a (maybe) SMBus problem, per below. Right now I've been up 6d 7h on an Athlon-optimized 2.4.4, and I think I had been up a similar amount of trouble free uptime when I had to shut it down due to thunderstorms last week. FWIW, I'm also running a 1.2, but with768MB of PC133 rather than DDR. The swap provided an instant cure for a hole pile of annoying problems that killed my system every few hours on the 8KTA3: crashes, hard hangs, unrequested GNOME session exits, and states where almost any command I typed resulted in a segfault. All that appears to be gone now. Re the (maybe) SMBus problem: I installed Lm_sensors to watch my temperatures, and had to kluge it a bit to get it to work. In particular, I had to load the i2c-ali1535 module rather than the i2c-ali15x3 module that Lm_sensors decided I needed. (I did this on the basis of the chip<==>module associations documented on the Lm_sensors page, ass-u-me-ing that the A7A266's M1535D+ should use the module specified for the M1535D.) However I now see this in /var/log/messages every few minutes: May 29 04:39:28 pollux kernel: i2c-ali1535.o: Resetting entire SMB Bus to clear busy condition (08) May 29 04:39:28 pollux kernel: i2c-ali1535.o: SMBus reset failed! (0x08) - controller or device on bus is probably hung I assume that the problem is because the module isn't quite working right for the D+ chip, but it does at least get my temperatures and voltages, and I haven't noticed any ill side effects other than the messages. Maybe it will go away when a new version of Lm_sensors comes out. > Perhaps Linux is not optimized > for performance with this chipset, but it feels fast to me. > I actually saw a consistent 2.7% speedup on a FP-intensive benchmark that my research group uses, comparing the A7A266 to the 8KTA3 after the switch, using the same hardware for everything else and with the BIOS settings matched as closely as possible. > According to 'hdparm -t /dev/hda', I am getting 25MB/s transfer rates > with my Quantum Fireball Plus LM. Seems a little high, but drive > performance 'feels' good. > I was getting around 32MB/s on my 8KTA3, but it dropped way down to 2.24 MB/sec after the upgrade. The DMA is off, but I don't want to just turn it on without investigating a possibly related problem first. Re the (maybe) DMA problem: I notice this whenever I boot, which may or may not all be relevant: May 22 21:45:07 pollux kernel: ALI15X3: IDE controller on PCI bus 00 dev 20 May 22 21:45:07 pollux kernel: PCI: No IRQ known for interrupt pin A of device 00:04.0. Please try using pci=biosirq. May 22 21:45:07 pollux kernel: ALI15X3: chipset revision 196 May 22 21:45:07 pollux kernel: ALI15X3: not 100%% native mode: will probe irqs later May 22 21:45:07 pollux kernel: ALI15X3: simplex device: DMA disabled May 22 21:45:07 pollux kernel: ide0: ALI15X3 Bus-Master DMA disabled (BIOS) May 22 21:45:07 pollux kernel: ALI15X3: simplex device: DMA disabled May 22 21:45:07 pollux kernel: ide1: ALI15X3 Bus-Master DMA disabled (BIOS) ... May 22 21:45:07 pollux kernel: PCI: Enabling device 00:0d.0 ( -> 0001) May 22 21:45:07 pollux kernel: PCI: Assigned IRQ 5 for device 00:0d.0 May 22 21:45:07 pollux kernel: Redundant entry in serial pci_table. Please send the output of May 22 21:45:07 pollux kernel: lspci -vv, this message (4793,4104,4793,215) May 22 21:45:07 pollux kernel: and the manufacturer and name of serial board or modem board May 22 21:45:07 pollux kernel: to [EMAIL PROTECTED] I did send the information to serial-pci-info. Notice that the kernel thinks it is talking to an ALI15X3, whereas my manual seems to say that I actually have an M1535D+ instead. While trying to discover a suitable IRQ for device 00:04.0, and noticed this: % /sbin/lspci -vv ... 00:04.0 IDE interface: Acer Laboratories Inc. [ALi] M5229 IDE (rev c4) (prog-if fa) Subsystem: Asustek Computer, Inc.: Unknown device 8053 Control: I/O+ Mem- BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B- Status: Cap+ 66Mhz- UDF- FastB2B+ ParErr- DEVSEL=medium >TAbort- SERR- Based on my weekend experience with this board and Linux, I think
Re: Status of ALi MAGiK 1 support in 2.4.?
On Mon, May 28, 2001 at 05:57:12PM +0200, Axel Thimm wrote: > What is the status of the support for this chipset, found for example in an > ASUS A7A266? Judging from > http://www.acerlabs.com/eng/support/faqlnx.htm > one gets the impression that ALi is respectfully treating the Linux community. I cannot answer your question about the level of support this chipset has, but suffice it to say that my new (as of last week) A7A266 based system (1.2Ghz T-Bird w/256MB Crucial DDR RAM) is running 2.4.5 (and previous to that 2.2.18) quite nicely. Perhaps Linux is not optimized for performance with this chipset, but it feels fast to me. According to 'hdparm -t /dev/hda', I am getting 25MB/s transfer rates with my Quantum Fireball Plus LM. Seems a little high, but drive performance 'feels' good. Based on my weekend experience with this board and Linux, I think I have made the right choice. - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Status of ALi MAGiK 1 support in 2.4.?
What is the status of the support for this chipset, found for example in an ASUS A7A266? Judging from http://www.acerlabs.com/eng/support/faqlnx.htm one gets the impression that ALi is respectfully treating the Linux community. Although VIA is the most popular chipset vendor, its reaction on the recently Southbridge 686 bugs (and even more recently the Northbridge bug with Athlon C) and the pirq routing problem in Linux rather disqualify it. So it seems that only AMD and ALi are left with Socket A chipsets (SiS 735 is yet to come). And the ALi MAGiK 1 has a good feature/cost ratio and is being reported as very stable (although perhaps not very fast). So having been burned in the recent past with VIAs KT133A/686B and Linux, I hope that I can get an ALi MAGiK 1 board and rest in peace ... Thanks, Axel. -- [EMAIL PROTECTED] - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
