Re: [Cooker] Frustration
On 19 Mar 2002, Pixel wrote: > FYI, you submitted your bug report on the 16 march. The 2 people (flepied and > fpons) that could help you on this are pretty busy/tired. And cooker has been > fairly noisy those days. You may have better luck mailing them directly, and > choosing a better date to post bugs (when in deep freeze only big problems are > taken into account others are devnulled) I submitted bug reports on 2 March, 6 March, and 9 March. This was during the 8.2 beta test period. I posted my reports here after trying (unsuccessfully) bugzilla, mandrakeforum, mandrakeexpert. The bug prevented the install kernel from booting (or, if one supplied the pci=conf2 argument, the kernel would boot, but the installer died at the second stage). I suggested the fix (remove the buggy ali15x3 driver from the instal kernel, since it provides no features necessary for installation and the generic IDE support works fine). The problem goes back as far as at least Mandrake 7.2 in my informal testing. I wanted to help get this resolved before 8.2 went gold. No responses at all, except from one user who responded off-list to tell me he also was having problems getting noticed by mdk maintainers. I was a paying user of Mandrake since 7.1; I've switched to debian on my laptop this week because this bug prevents mdk from installing, and I don't have a spare linux machine on which I can build a custom installer boot disk (and I won't bother now that I've got a nice, working system on my laptop - I didn't buy the thing to tinker with) If Mandrake is still around to put out another version, I might switch back. If the installer boots. -Isaac
[Cooker] Please!!! install kernel problem (ali15x3 hang) and solution
OK all, I'm trying to help here. I'm giving a clear description of a problem, the hardware it affects, and the proposed solution. PROBLEM: Installer kernel hangs at IDE probe when the ali15x3 driver is loaded and attempts to probe devices. The last few lines of output are: [...etc...] ALI15X3: IDE controller on PCI bus 00 dev 78 PCI: No IRQ known for interrupt pin A of device 00:0f.0. Please try using pci=biosirq. ALI15X3: chipset revision 195 ALI15X3: not 100% native mode: will probe irqs later That's it, it hangs dead at this point where it would normally probe the ide devices. The alt1 kernel which uses generic, non-dma IDEPCI does boot but does not support kernel pcmcia or any filesystem but ext2. :( Kernel options pci=biosirq and ide=nodma do nothing to prevent this lockup. HARDWARE: Fujitsu Lifebook P-2040, which uses an ALi 1533 PCI bridge and ALi 5229 IDE controller. SOLUTION PROPOSED: Unset CONFIG_BLK_DEV_ALI15X3=y (and CONFIG_WDC_ALI15X3=y, if configured) in the .config for the install kernel and recompile it without ali15x3 support. REASONING: The features of the ali15x3 driver are not necessary to the install process, and the alt1 kernel is not sufficiently feature complete for laptop installations (i.e. no kernel pcmcia support, no journaling or encrypted filesystems supported). Here's the part where I complain about the Mandrake bug reporting process: This is my third report of this bug here - I haven't recieved a reply or response or any acknowledgment at all, either privately or on the list. Please note that I did not come here first - I tried to jump through the bugzilla hoops, the mandrakeexpert hoops, and even the mandrakeforum hoops - my bugzilla account can't post new bugs (I did search for dupes and found none), mandrakeexpert won't let me post a question to mdk employees (despite the US$80 donation I made when I downloaded 8.1 instead of purchasing a retail box), and I never recieved my mandrakeforum password when I tried to register (twice!). Please restore my faith in this organization by at least replying to this message, even if only to say " off!" - that would be better than silence. Thanks, -Isaac
[Cooker] 8.2 beta 4 install kernel boot failure! ALI1533 driver hangs.
Just got my 8.2 beta 4 CDs burnt and fired up my Fujitsu Lifebook P-2040. I get the same result booting the default installer kernel and the alt2 (2.2.19) kernel from the second disk which is a hang like so: [...etc...] ALI15X3: IDE controller on PCI bus 00 dev 78 PCI: No IRQ known for interrupt pin A of device 00:0f.0. Please try using pci=biosirq. ALI15X3: chipset revision 195 ALI15X3: not 100% native mode: will probe irqs later That's it, it hangs dead at this point where it would normally probe the ide devices. The alt1 kernel which uses generic, non-dma PCI_IDE does boot but does not support kernel pcmcia or any filesystem but ext2. :( Kernel options pci=biosirq and ide=nodma do nothing to prevent this lockup. Is there any way we can get this driver (ali15x3.c) out of the installer kernel at least? I'm not aware of anything in the installation process that relies on it (the alt1 kernel doesn't have it, and it works just fine). Generic PCI IDE works, it's just the ali-specific driver that is locking it up. -Isaac
[Cooker] Buggy ALi 15x3 driver breaks some machines - remove from installkernel?
Hi all, This is my first post to Cooker, though I've been a (paying!) mdk user since 7.0. I'm just posting here to raise the issue of the ALi 15x3 kernel driver which apparently breaks some machines, including my own new Fujitsu Lifebook P-2040. The symptom is a hard hang during the kernel boot process, as others have reported. I am just raising the issue because this driver, while nice for DMA if it works on your machine, is completely unnecessary for the install process. Booting an alternate kernel off the second disk that does not have this driver works fine - but the alternate kernels only support ext2, which is unacceptable for portable machines with large disks. (At least this is true of Mdk8.1) So, all I ask is that this driver *not* be compiled into the main installer kernel of Mdk8.2-final. This will save those who are affected by the bugs in the /usr/src/linux/drivers/ide/alim15x3.c driver from having to install an ext2 system, and won't affect others at all. Is this possible? Thanks, -Isaac
