:> call VOP_GETPAGES() on a block of pages instead of just one. I'm
:> guessing that is not being done because it's about 100 lines of code
:> to do it right. It's easier just to call vn_rdwr() and let the
:> system do the clustering.
:
:This argument seems to apply to exec_map_first_pages() too. It does
:its own blocking to get up to VM_INITIAL_PAGEIN (normally 16) pages
:using only about 50 lines of code, but does this slightly wrong. It
:doesn't honor the device limit of si_iosize_max, unlike filesystem
:code. This used to break exec on zip drives (si_iosize_max used to
:be 32K, but VM_INITIAL_PAGEIN pages is 64K bytes on i386's).
:
:Bruce
Yes, it definitely applies. The vm_page_grab() can be replaced
with a vm_page_lookup() sequence similar to what we find in
do_sendfile(), and everything inside the 'if ((ma[0]->valid & ...'
can be ripped out and replaced with a single vn_rdwr() call,
if (error) test, and then loop back up to the vm_page_lookup().
There are enough parallels between the sendfile code and
the exec_map_first_page() code that a significant portion of
both procedures could be shifted to its own procedure which does
the read-ahead and returns a single, valid, busied page.
I am not volunteering to do this, though, I already have a full plate.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
