cvs commit: src/sys/conf files.ia64 src/sys/ia64/ia64 pmap.cuma_machdep.c

[Alan L. Cox]

Marcel Moolenaar wrote:
> 
> marcel      2003/09/20 12:27:48 PDT
> 
>   FreeBSD src repository
> 
>   Modified files:
>     sys/conf             files.ia64
>     sys/ia64/ia64        pmap.c
>   Added files:
>     sys/ia64/ia64        uma_machdep.c
>   Log:
>   Move uma_small_alloc() and uma_small_free() to uma_machdep.c. These
>   functions reference UMA internals from <vm/uma_int.h>, which makes
>   them highly unwanted in non-UMA specific files.
> 
>   While here, prune the includes in pmap.c and use __FBSDID(). Move
>   the includes above the descriptive comment.
> 
>   The copyright of uma_machdep.c is assigned to the project and can
>   be reassigned to the foundation if and when when such is preferrable.
> 
>   Revision  Changes    Path
>   1.59      +1 -0      src/sys/conf/files.ia64
>   1.123     +22 -77    src/sys/ia64/ia64/pmap.c
>   1.1       +82 -0     src/sys/ia64/ia64/uma_machdep.c (new)

An observation ... we now have three different locations, on four
different architectures, for these functions.  I definitely agree that
pmap.c is the wrong place because these functions have nothing to do
with page table and/or TLB management.  Long ago, I encouraged Jake to
start a trend for the better by placing the sparc64 implementations in
vm_machdep.c.  Whether it's uma_machdep.c or vm_machdep.c doesn't matter
much to me, only that we arrive at a consensus on which it is.  I would
observe the similarities to the new, optimized amd64 and ia64
sf_buf_alloc() implementations as an argument for vm_machdep.c.

In any case, I'll volunteer to update amd64 to match the new consensus.

Regards,
Alan