Thread (3 messages) flat view 3 messages, 2 authors, 2014-01-17

Re: Pull request: scottwood/linux.git

From: Scott Wood <hidden>
Date: 2014-01-17 23:00:42

On Wed, 2014-01-15 at 14:54 +1100, Benjamin Herrenschmidt wrote:
On Fri, 2014-01-10 at 18:44 -0600, Scott Wood wrote:
quoted
Highlights include 32-bit booke relocatable support, e6500 hardware
tablewalk support, various e500 SPE fixes, some new/revived boards, and
e6500 deeper idle and altivec powerdown modes.
This breaks WSP (A2) build with 64K pages:

/home/benh/linux-powerpc-test/arch/powerpc/mm/tlb_low_64e.S: Assembler messages:
/home/benh/linux-powerpc-test/arch/powerpc/mm/tlb_low_64e.S:334: Error: can't resolve `L0^A' {*ABS* section} - `PUD_SHIFT' {*UND* section}
/home/benh/linux-powerpc-test/arch/powerpc/mm/tlb_low_64e.S:334: Error: expression too complex
/home/benh/linux-powerpc-test/arch/powerpc/mm/tlb_low_64e.S:334: Error: operand out of range (67 is not between 0 and 63)
make[2]: *** [arch/powerpc/mm/tlb_low_64e.o] Error 1

I'm merging anyway because nobody uses WSP anymore (I'm keen to remove it by 3.15 or so)
Does this apply to all non-FSL 64-bit book3e, or is there anything else
that is still in use?
but in the meantime you may want to fix it (probably just ifdef the PUD level walk on
64k pages, look at what I do elsewhere).
Ifdeffing the PUD level doesn't make much sense in this case (other than
perhaps as the least intrusive way to make the error go away) because
that codepath is specific to e6500 which doesn't support 64K pages.  The
whole e6500 section should be ifdeffed out if there are 64K pages (or
really, if CONFIG_PPC_FSL_BOOK3E isn't defined, and 64K pages should
conflict with PPC_FSL_BOOK3E).

-Scott
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help