On Wed, 2016-01-20 at 13:23 +0100, David Kuehling wrote: > > > > > > "James" == James Cowgill <james410@cowgill.org.uk> writes: > > On Sun, 2016-01-17 at 06:39 +0100, David Kuehling wrote: > > > long time ago I was told that I need a newer PMON to boot recent > > > linux kernels [5], but then read on debian-mips about problems after > > > PMON upgrades [1]. > [..] > > > So the old PMON in those machines (only itx-a1101) has a bug where it > > loads the kernel at an address to close to the address the bootloader > > is stored at. If the kernel being loaded has a large bss section (or > > is large in memory for any reason), PMON will overwrite itself and > > crash. > > > The bug appeared after this kernel commit: > > https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=c46173183657bbdbe0d54a981c28807581648422which > > increased the size of the bss section to something like 20M on all > > Loongson 3 kernels. > > > Back then I tried using PMON 4.something but it wouldn't boot any > > kernel I threw at it. > > > In the end I patched the kernel to reduce the value of > > MAX_PHYSMEM_BITS from 48 to 35 and it worked again - I think this > > should be ok on all Loongson 3A machines. > > > Hopefully if you try that, any recent kernel will work with the old > > PMON. > > Ok, just to make sure I understood correctly: you never managed to make > the newer pmon 4 work with any kernels at all? That would be somewhat > surprising, given that people were repeatedly told to upgrade PMON on > these Loongson3 boxes. Yes you understood correctly, none of the small number of kernels I tried would boot with PMON 4. Since I could never find the definitive source and since the other solution was "good enough" at the time, I never looked any further into it. Maybe there was a new PMON version in the meantime? Like I said this was all some time ago. James
Attachment:
signature.asc
Description: This is a digitally signed message part