Hi,
I have a very old Pentium-1 PC (60 MHz).
I have installed 4.11, without any problems,
although the dmesg output has a line about buggy
ATA controller (see below for dmesg output).
When I try to install 5.3 in this PC (using the
three floppies), I get to the point where FDISK
should start, but instead I get an error message
that says something like "No disk is found".
Any ideas why 4.11 can handle this, but 5.3 cannot?
Moreover, 4.11 seems to be running fine on this PC.
Regards,
Rob.
------------------------------------------------
'dmesg' output after 4.11 installation:
CPU: Pentium/P5 (58.18-MHz 586-class CPU)
Origin = "GenuineIntel" Id = 0x517 Stepping = 7
Features=0x1bf<FPU,VME,DE,PSE,TSC,MSR,MCE,CX8>
real memory = 25165824 (24576K bytes)
avail memory = 19152896 (18704K bytes)
Preloaded elf kernel "kernel" at 0xc055c000.
Intel Pentium detected, installing workaround for
F00F bug
md0: Malloc disk
npx0: <math processor> on motherboard
npx0: INT 16 interface
pcib0: <Host to PCI bridge> on motherboard
pci0: <PCI bus> on pcib0
atapci0: <RZ 100? ATA controller !WARNING! buggy
chip data loss possible> at device 1.0 on pci0
atapci0: ATA channel disabled by BIOS
[...cut...]
ad0: 520MB <ST3660A>
[1057/16/63] at ata0-master BIOSPIO
ad1: 2423MB <SAMSUNG WINNER-3 WN32543A(2.5GB)>
[4924/16/63] at ata0-slave BIOSPIO
__________________________________
Do you Yahoo!?
Yahoo! Small Business - Try our new resources site!
http://smallbusiness.yahoo.com/resources/
Torfinn Ingolfsen
2005-Mar-27 02:01 UTC
buggy ATA controller: I can install 4.11, but not 5.3 !?!
On Sun, 27 Mar 2005 00:10:09 -0800 (PST) Rob <spamrefuse@yahoo.com> wrote:> I have installed 4.11, without any problems, > although the dmesg output has a line about buggy > ATA controller (see below for dmesg output).The controller is broken. For the whole ugly story, see http://mindprod.com/eideflaw.html> When I try to install 5.3 in this PC (using theWhat does the dmesg say, if yor press "Scroll Lock" and page up in it? Is your ata controller and / or disk detected at all?> Moreover, 4.11 seems to be running fine on this PC.Well, if it works, why don't you just keep it? -- Regards, Torfinn Ingolfsen, Norway
Torfinn Ingolfsen wrote:> On Sun, 27 Mar 2005 00:10:09 -0800 (PST) > Rob <spamrefuse@yahoo.com> wrote: > >>I have installed 4.11, without any problems, >>although the dmesg output has a line about buggy >>ATA controller (see below for dmesg output). > > > The controller is broken. For the whole ugly story, > see http://mindprod.com/eideflaw.htmlYes, indeed. But I knew about that.>>When I try to install 5.3 in this PC (using the > > What does the dmesg say, if yor press "Scroll Lock" > and page up in it? > Is your ata controller and / or disk detected at > all?Well, I got stuck during the initial stage of the install process; I haven't got any kernel info at that point, or have I?>>Moreover, 4.11 seems to be running fine on this PC. > > Well, if it works, why don't you just keep it?That's not the issue here. Of course I stick with 4.11; I don't have a choice. With this email, I want to report that this problem is 'solved' by 4.11 and prior, but not anymore in 5.3. Rob. __________________________________ Do you Yahoo!? Yahoo! Mail - You care about security. So do we. http://promotions.yahoo.com/new_mail
--- Randy Rowe <rerowe@rerowe.com> wrote:> On Sun, 2005-03-27 at 00:10 -0800, Rob wrote: > > > > I have a very old Pentium-1 PC (60 MHz). > > > > I have installed 4.11, without any problems, > > although the dmesg output has a line about buggy > > ATA controller (see below for dmesg output). > > ------------------------------------------------ > > > > 'dmesg' output after 4.11 installation: > > > > CPU: Pentium/P5 (58.18-MHz 586-class CPU) > > atapci0: <RZ 100? ATA controller !WARNING! buggy > > chip data loss possible> at device 1.0 on pci0 > > atapci0: ATA channel disabled by BIOS > > If the above line is true, I am surprised that 4.11 > works. > Is the channel really disabled in BIOS?First of all, I'm not expert in these matters. I'm just curious why I can install 4.11, but not 5.3. (I wish I could install 5.3 !!). Indeed, the BIOS is a strange issue. In the BIOS setup, I can select 1) Standard CHS 2) Logical block 3) Extended CHS, 4) Auto Detected But whatever I select here, when I get back into the the BIOS setup, to verify check my settings, I find that 1) is selected. Apparently this is what means "disabled by BIOS", but I'm not sure. The 4.11 dmesg output has following: ------------------------------------------------ CPU: Pentium/P5 (55.18-MHz 586-class CPU) Origin = "GenuineIntel" Id = 0x517 Stepping = 7 Features=0x1bf<FPU,VME,DE,PSE,TSC,MSR,MCE,CX8> -//- atapci0: <RZ 100? ATA controller !WARNING! buggy chip data loss possible> at device 1.0 on pci0 atapci0: ATA channel disabled by BIOS -//- ata0 at port 0x1f0-0x1f7,0x3f6 irq 14 on isa0 ata1 at port 0x170-0x177,0x376 irq 15 on isa0 -//- ad0: 520MB <ST3660A> [1057/16/63] at ata0-master BIOSPIO ------------------------------------------------ Notice that the harddisks are on isa0, not on atapci0; probably because of this buggy chip and/or BIOS problem. Could it be that 5.3 install cannot (yet?) handle harddisks on isa properly? Or must I help 5.3 by setting some hints/syscontrols before loading the kernel? Thanks, Rob. __________________________________ Do you Yahoo!? Yahoo! Small Business - Try our new resources site! http://smallbusiness.yahoo.com/resources/
--- Mars Trading <marstrade@gmail.com> wrote:> Rob, > > I think you may be on to something here. Not that > I'm an expert, BTW. > If I remember correctly, 4.11 doesn't use > device.hints; > device irq's and stuff were all included in the > kernel configuration. > > This is no longer the case with 5.x which uses > /boot/device.hints to tell where attached devices > are. But how does one edit device.hints on a > boot cd?I found a webpage: http://www.gsp.com/cgi-bin/man.cgi?section=4&topic=acd which mentions a bit about syscontrols, that might be relevant to my problem. So I hope this can be solved with 5.3 by setting the proper hints/syscontrols as loader settings before the kernel is loaded. But I'm don't know whether the 5.3 install process allows me doing this. How can I modify the install process and which hints/controls should I modify? I hope someone here on the list has some good ideas :). BTW: I'm not booting from CD, but from the three floppies, followed by a network install; but that shouldn't matter. Regards, Rob. __________________________________ Do you Yahoo!? Yahoo! Small Business - Try our new resources site! http://smallbusiness.yahoo.com/resources/
Rob wrote:> > I have already put this machine under moderate load > and recompiled/installed a new world/kernel without > any problems. Apparently 4.11 knows how to bypass > the flaws of this buggy ATA controller. At least > that's my impression. > Would 5.3 or 5.4 do this as-good here?Googling on this topic, I found info on a Linux kernel configuration site: <Quote> CONFIG_BLK_DEV_RZ1000 The PC-Technologies RZ1000 chip is used on many common 486 and Pentium motherboards, usually along with the "Neptune" chipset. Unfortunately, it has a rather nasty design flaw that can cause severe data corruption under many conditions. Say Y here to include code which automatically detects and corrects the problem under Linux. This may slow disk throughput by a few percent, but at least things will operate 100% reliably. If unsure, say Y. <\Quote> I wonder whether FreeBSD (at least 4.X) uses the same strategy to bypass the RZ1000 trouble on my PC. Regards, Rob. __________________________________ Do you Yahoo!? Take Yahoo! Mail with you! Get it on your mobile phone. http://mobile.yahoo.com/maildemo
--- Doug White <dwhite@gumbysoft.com> wrote:> > AFAIK the RZ1000 bug was that it would corrupt data > to the slave channel if the primary channel was > also active. If you only have one device then > you may not be able to reproduce it.I have two harddisks on ata0: ad0: 520MB <ST3660A> [1057/16/63] at ata0-master BIOSPIO ad1: 2423MB <SAMSUNG WINNER-3 WN32543A(2.5GB)> [4924/16/63] at ata0-slave BIOSPIO ad0 has the base OS, and ad1 has /usr/src and /usr/obj for recompiling world and kernel. So far no problems. However, if I remember well, the other IDE connector on the motherboard does not seem to work, which now I realize could be caused by the buggy RZ 1000 chip.> Do you have verbose boot output from the non-working > 5.x boot?No, not at the moment. I may try 5.3 (or probably 5.4) later once again. What part of the output would be particularly interesting? I ask, because I am managing this PC at the other end of the world, giving instructions to a non-Unix, non-FreeBSD user overthere :). Regards, Rob. __________________________________ Do you Yahoo!? Yahoo! Sports - Sign up for Fantasy Baseball. http://baseball.fantasysports.yahoo.com/
--- Doug White <dwhite@gumbysoft.com> wrote:> On Thu, 31 Mar 2005, Rob wrote: > >> No, not at the moment. I may try 5.3 (or probably >> 5.4) later once again. What part of the output >> would be particularly interesting? >> I ask, because I am managing this PC at the other >> end of the world, giving instructions to a >> non-Unix, non-FreeBSD user overthere :). > > Ugh. That will make this really hard to debug then. > To get the verbose output you want to use a serial > console. That output will say why it won't attach > the ata controller.OK, I will certainly try that then, giving the proper insturctions, since this Pentium1 FreeBSD PC will soon have a Windows PC next to it. However, I need a little advice/help: the handbook is still out-of-date for making serial console install floppies. Chapter 2.12 of the handbook talks about kern.flp and mfsroot.flp, whereas we have three floppies with 5.X install. Which one(s) of the three floppies of 5.X needs to be modified by the procedure of chapter 2.12 ? "Windows HyperTerminal" is then the way to go, isn't it? Thanks, Rob. __________________________________ Yahoo! Messenger Show us what our next emoticon should look like. Join the fun. http://www.advision.webevents.yahoo.com/emoticontest
Doug White wrote:> On Thu, 31 Mar 2005, Rob wrote: > >>--- Doug White <dwhite@gumbysoft.com> wrote: >> >>>AFAIK the RZ1000 bug was that it would corrupt data >>>to the slave channel if the primary channel was >>>also active. If you only have one device then >>>you may not be able to reproduce it. >> >>I have two harddisks on ata0: >> >> ad0: 520MB <ST3660A> >> [1057/16/63] at ata0-master BIOSPIO >> ad1: 2423MB <SAMSUNG WINNER-3 WN32543A(2.5GB)> >> [4924/16/63] at ata0-slave BIOSPIO >> >>ad0 has the base OS, and ad1 has /usr/src and >>/usr/obj for recompiling world and kernel. >> >>So far no problems. >>However, if I remember well, the other IDE connector >>on the motherboard does not seem to work, which now >>I realize could be caused by the buggy RZ 1000 chip. > > Entirely possible.With FreeBSD 4-Stable running still fine, I get this: # grep ata /var/run/dmesg.boot atapci0: <RZ 100? ATA controller !WARNING! buggy chip data loss possible> at device 1.0 on pci0 atapci0: ATA channel disabled by BIOS ata0 at port 0x1f0-0x1f7,0x3f6 irq 14 on isa0 ata1 at port 0x170-0x177,0x376 irq 15 on isa0 ad0: 520MB <ST3660A> [1057/16/63] at ata0-master BIOSPIO ad1: 2423MB <SAMSUNG WINNER-3 WN32543A(2.5GB)> [4924/16/63] at ata0-slave BIOSPIO # vmstat -i | grep -i ata interrupt total rate rl0 irq11 3187504 1 fdc0 irq6 2 0 ata0 irq14 33429902 17 sio0 irq4 3 0 sio1 irq3 1 0 clk irq0 187773091 100 rtc irq8 240341556 127 Total 464732059 247 So ata1 (irq 15) is not in the interrupt table !?! What does that mean? As a reminder: I have following in dmesg output: atapci0: <RZ 100? ATA controller !WARNING! buggy chip data loss possible> at device 1.0 on pci0 atapci0: ATA channel disabled by BIOS For the harddisk to be recognized by the kernel, I need exactly this in the kernel config: device ata0 at isa? port IO_WD1 irq 14 device ata1 at isa? port IO_WD2 irq 15 Without these lines in the kernel config, I get a fatal 'no root device found' error at bootup. Apart from these peculiar findings, 4-Stable is running flawlessly on this old Pentium-1 PC. ----- Another reminder: when I try to install 5.3 on this PC, I get a "no disk found" error at a very early stage of the installation procedure. I hope to investigate this further in a not-so-near future.... Regards, Rob. __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com
--- Graham Menhennitt wrote:> Rob wrote: > > >Another reminder: when I try to install 5.3 on this > >PC, I get a "no disk found" error at a very early > >stage of the installation procedure. I hope to > >investigate this further in a not-so-near > future.... > > > > > Rob, > > I have a similar problem with 5.3 although I'm not > doing a new install - I'm upgrading from an earlier > 5.X. Can you please try installing a 5.X > from December last year or earlier. Actually, you > don't need to install anything - I'd just like to > know whether it can see the disk. I've > faised PR kern/79332 about it which contains two > possible fixes.Thank you. The computer is my brother's old Pentium-1 (60 MHz); he and the computer are at the other end of the world. Only when my brother has spare time, we open a live chat connection and I assist him with certain console tasks (he is not at all a Unix/FreeBSD person). However, most of the time I manage this PC remotely (it's now running 4-Stable flawlessly). At some point, we may try once again to install a 5.X version; it should be a straightforward install procedure, because too much freaky stuff might be too complex to explain to a non-Unix, non-FreeBSD person.... My plan for next time, is to use a serial console during install, so that I can receive the kernel output to get a clue why 5.X cannot find the disk, which 4-Stable does find easily. I may travel to my brother's home some time this year; then I will do whatever freaky tricks are needed to find out why 5.X is not installing properly. Regards, Rob. __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com