Re: Linux 2.4.0test11pre1ac1

[Jeff Garzik <jgarzik@mandrakesoft.com>]

the network driver changes look ok except for:
epic100:
* CARDBUS is never defined.  Should that be CONFIG_CARDBUS?
* just increment the version number.  There's no need to add "a" on the
end...  this version number just differentiates us from the 'canonical'
Donald Becker version of epic100.c.


net/atm/pvc:  return the error value from sock_register, not toss it
away.


ramfs comments:
* Hang on to the ramfs changes for a day or two, there is kmap cleanup
(now returns void*) going to Linus RSN.
* Does ramfs_statfs() need that sb lock in it?  Sure free_pages/inodes
might be getting updated on some other CPU, but its statfs so who
cares...
* the default ramfs maxsize, half of all RAM, seems a little
conservative.
* there is no need to kmalloc a private superblock structure, when room
is already allocated inside the superblock structure for private data. 
(sb->u)
* 


I wonder if we really need removepage added to struct address_space? 
That's one API change we shouldn't throw in without discussion, IMHO...
it screams "ramfs-specific hack in core code!"
 void __remove_inode_page(struct page *page)
 {
+       struct address_space *mapping = page->mapping;
+
+       if (mapping && mapping->a_ops && mapping->a_ops->removepage)
+               mapping->a_ops->removepage(page);
+


And finally, don't you need to EXPORT_SYMBOL pm_devs_lock ?



-- 
Jeff Garzik             |
Building 1024           | Would you like a Twinkie?
MandrakeSoft            |
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/