Unix code in Linux

Yes, it's ugly as hell.
As far as I can see, the only user of ate_malloc are a few rmalloc calls. There is one rmalloc_align call, but afaics the function is not implemented.

The code is filled with #defines that rename linux functions - mutex_spinlock, spin_lock_destroy(), whatever. There is so much renaming that they even create a prototype for a #define:

(it seems sn1 got killed recently, I searched in lxr.linux.no)

AFAICS ate_malloc should die, and the rmalloc callers should use request_resource & friends from <linux/ioport.h>.

--
Manfred

atealloc is even worse than I noticed immediately:

* it allocates a semaphore, just in case. Never used.
* it allocates a spinlock and stores the pointer to the spinlock in .m_size of the 2nd map entry.
* then defines and one-line wrapper functions to obfuscate what's going on:
pcibr_ate_alloc is a wrapper around rmalloc. rmalloc is a #define to atealloc
atealloc is the map allocator.
* it seems the author never heared of include files - the function prototypes are scattered around in the .c files, instead of including <asm/sn/ate_utils.h>

The actual users are in arch/ia64/sn/io/sn2/

There are 2 users:
*********
win_map: It seems to be used for locating io port and memory ranges for pci device initialization.

* creation: in pciio.c:
- new maps are allocated with pciio_device_win_map_new(). They are initially empty.
- The maps are filled with pciio_device_win_populated().
- they are destroyed with pciio_device_win_map_free().
* use:
- allocation with pciio_device_win_alloc().
- free with pciio_device_win_free().

Users are in pcibr/pcibr_{slot,dvr}.c: pcibr_bus_addr_alloc and pcibr_bus_addr_free.

*********
bs_ext_ate_map and bs_int_ate_map: management of address translation entries (ATE).

* creation: created inside pcibr_attach2, destroyed in pcrbr_detach. pcibr/pcibr_dvr.c
* use: in pcibr/pcibr_ate.c
pcibr_ate_alloc, pcibr_ate_free.

The real user is pcibr_dmamap_alloc.
*********

What about deleting ate_utils.c and adding a rmalloc implementation (textbook or bsd) to linux/lib?
Any volunteers?
--
Manfred
P.S.: I'd propose that the GPL rule that a change must be tagged with the name of the person who changes (or adds) a file is enforced - atealloc.c is only tagged with "SGI", thus I don't know who should be shot for writing that.