Re: [RFC][PATCH v3] powerpc: e300c2/c3/c4 TLB errata workaround

2009-03-16 Thread Kumar Gala
On Mar 13, 2009, at 5:06 PM, Kumar Gala wrote: On Mar 13, 2009, at 4:26 PM, Benjamin Herrenschmidt wrote: +BEGIN_FTR_SECTION + b TlbWo/* Code for TLB-errata workaround doesn't fit here */ +END_FTR_SECTION_IFSET(CPU_FTR_NEED_DTLB_SW_LRU) +RFTlbWo: Can you use nicer label names ?

Re: [RFC][PATCH v3] powerpc: e300c2/c3/c4 TLB errata workaround

2009-03-13 Thread Kumar Gala
On Mar 13, 2009, at 4:26 PM, Benjamin Herrenschmidt wrote: +BEGIN_FTR_SECTION + b TlbWo/* Code for TLB-errata workaround doesn't fit here */ +END_FTR_SECTION_IFSET(CPU_FTR_NEED_DTLB_SW_LRU) +RFTlbWo: Can you use nicer label names ? :-) Also, that's a lot of code for such a hot pa

Re: [RFC][PATCH v3] powerpc: e300c2/c3/c4 TLB errata workaround

2009-03-13 Thread Benjamin Herrenschmidt
> +BEGIN_FTR_SECTION > + b TlbWo/* Code for TLB-errata workaround doesn't fit here */ > +END_FTR_SECTION_IFSET(CPU_FTR_NEED_DTLB_SW_LRU) > +RFTlbWo: Can you use nicer label names ? :-) Also, that's a lot of code for such a hot path... Cheers, Ben. _

Re: [RFC][PATCH v3] powerpc: e300c2/c3/c4 TLB errata workaround

2009-03-13 Thread Benjamin Herrenschmidt
On Fri, 2009-03-13 at 10:16 -0500, Kumar Gala wrote: > From: David Jander > > Complete workaround for DTLB errata in e300c2/c3/c4 processors. > > Due to the bug, the hardware-implemented LRU algorythm always goes to way > 1 of the TLB. This fix implements the proposed software workaround in > fo

[RFC][PATCH v3] powerpc: e300c2/c3/c4 TLB errata workaround

2009-03-13 Thread Kumar Gala
From: David Jander Complete workaround for DTLB errata in e300c2/c3/c4 processors. Due to the bug, the hardware-implemented LRU algorythm always goes to way 1 of the TLB. This fix implements the proposed software workaround in form of a LRW table for chosing the TLB-way. Signed-off-by: David Ja