Re: CVS commit: src/sys/uvm

[yamt%mwd.biglobe.ne.jp@localhost (YAMAMOTO Takashi)]

[ moved from souce-changes@ ]

> YAMAMOTO Takashi wrote:
> >> Module Name:       src
> >> Committed By:      chris
> >> Date:              Sat Feb 23 17:27:58 UTC 2008
> >>
> >> Modified Files:
> >>    src/sys/uvm: uvm_km.c uvm_map.c uvm_page.c
> >>
> >> Log Message:
> >> Add some more missing pmap_update()s following pmap_kremove()s.
> >>
> >>
> >> To generate a diff of this commit:
> >> cvs rdiff -r1.97 -r1.98 src/sys/uvm/uvm_km.c
> >> cvs rdiff -r1.250 -r1.251 src/sys/uvm/uvm_map.c
> >> cvs rdiff -r1.128 -r1.129 src/sys/uvm/uvm_page.c
> >>
> >> Please note that diffs are not public domain; they are subject to the
> >> copyright notices on the relevant files.
> >>     
> >
> > why?
> > for kmem grovellers?
> >
> > YAMAMOTO Takashi
> >   
> 
> We document the pmap API as requiring it. Pmap can defer doing work to 
> finalize the changes in hardware until pmap_update is called, so not 
> doing so could cause issues.

at least, pmap_update calls you added to uvm_km_free and uvm_pagezerocheck
seem redundant to me.

YAMAMOTO Takashi

> 
> This was actually triggered by reviewing a patch of Joerg's and finding 
> a few places in the kernel that don't follow the documentation, so I 
> thought I'd fix them. See:
> http://mail-index.netbsd.org/tech-kern/2008/02/23/msg000442.html
> and it's follow-ups.
> 
> Thanks,
> Chris