- 1 Fedora SPARC
- 1.1 NEWS
- 1.2 Status
- 1.3 Goals and Objectives
- 1.4 Architecture Maintainer Team
- 1.5 Contact Info
- 1.6 History
- 1.7 SPARC LiNUX docs
- 1.8 CPU and Architecture Target
- 1.9 Sandbox Systems
- 1.10 Buildservers
- 1.11 Tracker Bug
- 1.12 Releases
- 1.13 Alternative distributions for SPARC
In October 2015, ORACLE released a SPARC-sun4v reference platform, for T1 or better(UA2005->), based on RHEL-6:
In early 2017, the SPARC version of Oracle Linux 6.7 was released for T4, T5, M5, M6, T7 and S7 platforms(OSA2011->):
It also runs in Oracle VM Server for SPARC(Logical Domains):
The development of this architecture is halted (mailing list thread) as of 2012 because lack of manpower and/or interest in the community. Latest release is 12-BETA.
Goals and Objectives
The primary goal of this project is to provide support for SPARC as a secondary architecture in Fedora.
Architecture Maintainer Team
- Tom "spot" Callaway (team lead) (Fedora Account: spot)
- Dennis Gilmore (Fedora Account: ausil)
- Patrick "Jima" Laughton (Fedora Account: jima)
- Peter Jones (Fedora Account: pjones)
Interested in joining the SPARC team? Contact the team lead.
- IRC: freenode.net
- Regular IRC Meeting: Fedora SPARC will have regular IRC meetings on TBD.
- Mailing List:
Fedora SPARC started out as the Aurora SPARC Linux Project, when Red Hat stopped support for SPARC with Red Hat Linux 6.2. Fedora SPARC is the continuation of the Aurora efforts, in an official capacity as part of the Fedora Project.
SPARC LiNUX docs
CPU and Architecture Target
The baseline SPARC CPU architecture that we have chosen to support is sparcv9 in 32bit. Although Aurora (the predecessor to Fedora SPARC) supported sparcv7 as its baseline, poor upstream kernel support for sparc(32-bit) combined with significant optimization gain make sparcv9 a much more viable target. Unfortunately, this means that sparc32 systems (anything sun4c, sun4d, sun4m and LEON) will not work with Fedora SPARC.
It is possible that we will release a sparcv7 baselined tree again at some point, either under the Aurora name or as a Fedora SPARC release.
Currently, there are no Fedora SPARC sandbox systems available.
|daedalus||Sun Fire T1000, 32 1ghz thread Niagara 1 CPU, 2x73gb sas drives, 16gb ram.|
|korolev||Sun Fire T2000, 32 1ghz thread Niagara 1 CPU, 2x73gb sas drives, 8gb ram, 253gb T3 StorEdge shelf.|
|vala||Netra X1, 500mhz Ultrasparc IIi, 2x80gb ide drives, 2gb ram.|
Build results mailing lists
- Mailing List:
If excluding sparc architectures you need to make the bug block F-ExcludeArch-SPARC
to see whats currently blocking visit Bugzilla
Fedora SPARC 12 BETA
- ISOs are available here: http://archive.fedoraproject.org/pub/archive/fedora-secondary/releases/test/12-Beta/sparc/iso/
- FILES are available here: http://archive.fedoraproject.org/pub/archive/fedora-secondary/releases/test/12-Beta/
- UPDATES are available here: http://archive.fedoraproject.org/pub/archive/fedora-secondary/updates/12/
- SBUS devices do not autodetect
- serial console install - By default the serial console is not enabled. To install with only a serial console you just boot into single user mode with a serial console by adding "console=ttyS0,9600 single" to the SILO boot prompt, At the '# ' prompt edit /boot/silo.conf to remove the rhgb and add "console=ttyS0,9600". Then copy /etc/event.d/tty1 /etc/event.d/serconsole and edit serconsole to use ttyS0 in place of tty1. You then must edit /etc/securetty to include ttyS0 for pam to accept the root password on the serial console. After doing this 'exit' will boot multi-user and you should see a login prompt on the serial console. If you do not do these steps, the multi-user boot will appear to hang.
- serial console install - to do any customisation at all you need to use vnc
- sunblade and Fire systems with qlogic hba's for primary storage - Make sure that you do a graphical install (vnc or local X) or kickstart and include "hardware support" this is so that you have the firmware for your primary storage
- install blowing up at partitioning - parted doesnt seem to like old parted disk labels. in rescue mode "dd if=/dev/zero of=/dev/sdX bs=512 count=2" where X == the value of each disk (e.g. /dev/sda)
- partitioning is fragile - simple partitioning works fine more complicated setups might require you to manually use parted to setup the partitioning you want and then choosing custom partitiing and allocate as you need to
- systems with less than 512mb ram - using vnc you may get OOM during package install. you will need to either do a default install or select a smaller package set
- repo issues when installing - There repos are not laid out exactly as anaconda wants when doing a netinstall. you can pass repo="path to mirror" or fix up in the gui