This is the list of known errata for the 0.9.3 Release.
See the previous erratas as well.
If you select DHCP network configuration during install, and the system fails to configure itself, it will not tell you. Instead, it waits until its requests time out and then claims the configuration was successful. If you have doubts, scroll down the menu to execute a shell, and then use 'ifconfig -a' to check the network state. Your ethernet port will have an address listed (inet addr) if the DHCP was successful. If not, reselect Configure Network from the menu to try again.
When installing on systems that support both 32/64(like B/C/JXX00) palo will initially boot a 32bit kernel. If you select a 64bit kernel durning the install, the install will not be able to complete the "Configuring Kernel Modules" section because it cannot find any modules. You can safely skip over this section.
For systems which support STI Framebuffer, the X Windows
System does not configure correctly upon initial install.
There are two issues. First, the config file for the xserver-xfree86 package is missing a clause for hppa, which prevents easy configuration of the X server. To fix this, either
apply the following patch to your /var/lib/dpkg/info/xserver-xfree86.config file, and then run "dpkg-reconfigure xserver-xfree86"
upgrade to the 4.1.0-11 or newer version of the X packages, available in unstable, which fixes the problem
The second issue is that the screen resolution you choose when configuring X must match the screen resolution set in the firmware and discovered by the kernel. The best approach is to interrupt the boot and make sure the firmware settings for your monitor's resolution and frame refresh rate are satisfactory, then select only that resolution when answering the X server configuration questions. You may get the real resolution of your screen by running dmesg | grep "^fb" or by running fbset -i (as root).
You could also choose to edit your /etc/X11/XF86Config-4 file by hand if you like that sort of thing...
"I'm not trying to set up the PCI graphic card now... It would freeze your machine"
During the initialisation of the A4977A Visualize EG PCI card
on C3000 models (and we assume J5000), the machine completely
freezes. However, the same card works fine in most other
machines. Newer (2.4.16 based) kernels checks for the broken
combination and avoids it. Thus, owners of older machines
may get a 2.4.16-pa12 (or newer) kernel from our
The internal IDE CDROM on C3000 and J5000, at least, is not currently supported. An external SCSI CDROM can be used, but for that to work you must interact with IPL and add an extra parameter "sym53c8xx=buschk:2". Extra parameters are added simply by selecting an existing parameter and adding to the end of it.
Filesystem Corruption! Systems with NCR/LSI 53c720
Fast/Wide/Differential SCSI interface need to use
"ncr53c8xx=burst:3" palo boot parameter. This avoids
overflowing the on-chip DMA FIFO which results in file
system corruption on disks behind that interface.
Either edit /etc/palo.conf or add this manually at boot time.
2.4.16-pa14 (and newer) kernels with a "real" fix are
available on our
In order to use ISA cards (e.g. an old ISA NE2000 clone) in your EISA slots you need to add the extra palo parameter "eisa_irq_edge=3,4,5,7,9,10,11,14,15". The numbers define the IRQs which should be available by ISA.