Linux 6.10-rc2 Will Fix Booting Intel Quark CPUs & Problems With Old AMD CPUs
Ahead of today's Linux 6.10-rc2 kerne weekly test release a few "x86/urgent" patches were submitted for addressing some fallout on Intel and AMD processors.
One of the x86 urgent fixes this week is to restore support for booting Linux on the Intel Quark SoCs... The 32-bit SoCs that held the distinction of being even slower than Atom while enjoying very little power consumption. The Quark X1000 from a decade ago had a 400MHz clock speed for its single core, DDR3-800 memory support, and a TDP of just 2.2 Watts.
The ability to boot Linux on the Intel Quark and some other processors regressed last year. Due to an unintended change last year, bad things will happen in the kernel if the default cache line size can not be enumerated. The change marked for Linux 6.10-rc2 and to be back-ported to stable kernel versions is to assume a default cache line size of 32 (rather than 0) if it can't otherwise be enumerated by the processor.
Over on the AMD side, this week's x86 urgent pull happens to fix some topology parsing issues for AMD Family 15h (Bulldozer) and older processors. A botched check on evaluating SMT support had messed up the support for pre-Zen AMD processors.
This week's x86 fixes are now waiting to be pulled ahead of the Linux 6.10-rc2 kernel release later today.
One of the x86 urgent fixes this week is to restore support for booting Linux on the Intel Quark SoCs... The 32-bit SoCs that held the distinction of being even slower than Atom while enjoying very little power consumption. The Quark X1000 from a decade ago had a 400MHz clock speed for its single core, DDR3-800 memory support, and a TDP of just 2.2 Watts.
The ability to boot Linux on the Intel Quark and some other processors regressed last year. Due to an unintended change last year, bad things will happen in the kernel if the default cache line size can not be enumerated. The change marked for Linux 6.10-rc2 and to be back-ported to stable kernel versions is to assume a default cache line size of 32 (rather than 0) if it can't otherwise be enumerated by the processor.
Over on the AMD side, this week's x86 urgent pull happens to fix some topology parsing issues for AMD Family 15h (Bulldozer) and older processors. A botched check on evaluating SMT support had messed up the support for pre-Zen AMD processors.
This week's x86 fixes are now waiting to be pulled ahead of the Linux 6.10-rc2 kernel release later today.
4 Comments