top stack users for 2.5.71

[Jörn Engel]

44 functions for 2.5.67.
45 functions for 2.5.68
41 functions for 2.5.69
36 functions for 2.5.70
32 functions for 2.5.71

Five less, one more.  The newcomer this time is in
drivers/usb/storage/isd200.c:
0xc0b9c9a0 isd200_action:                                sub    $0x444,%esp

This is curious, since no obvious change causing this is in the 2.5.71
patch.

The two big consumers on the stack are these two, that use a little
over 256k and a little over 512k respectively.
        struct scsi_cmnd srb;
        struct scsi_device srb_dev;

This one really puzzles me, as those two variables were already present
in 2.5.67.  A closer look reveals that the function isn't exactly new,
just oszillating around the 1k limit.

2.5.63:
0xc08c0be3 <isd200_action+3/1ac>:                        sub	$0x16c,%esp
2.5.64 - 2.5.67: <1k
2.5.68:
0xc0b73c60 isd200_action:                                sub	$0x424,%esp
2.5.69:
0xc0b4bea0 isd200_action:                                sub	$0x414,%esp
2.5.70: <1k
2.5.71:
0xc0b9c9a0 isd200_action:                                sub    $0x444,%esp

Quite an interesting behaviour that I don't fully understand yet.

P 0xc022f0d6 presto_get_fileid:                            sub    $0x119c,%esp
P 0xc022d896 presto_copy_kml_tail:                         sub    $0x1028,%esp
0xc0929258 ide_unregister:                               sub    $0x96c,%esp
0xc0e7f4d3 snd_emu10k1_fx8010_ioctl:                     sub    $0x830,%esp
0xc0876936 w9966_v4l_read:                               sub    $0x828,%esp
0xc0e0fb0b snd_cmipci_ac3_copy:                          sub    $0x7c0,%esp
0xc0e1012b snd_cmipci_ac3_silence:                       sub    $0x7c0,%esp
0xc0105650 huft_build:                                   sub    $0x59c,%esp
0xc01073d0 huft_build:                                   sub    $0x59c,%esp
0xc02df9b6 dohash:                                       sub    $0x594,%esp
0xc0108256 inflate_dynamic:                              sub    $0x554,%esp
0xc01064a6 inflate_dynamic:                              sub    $0x538,%esp
0xc0223996 presto_ioctl:                                 sub    $0x508,%esp
0xc0e797a8 snd_emu10k1_add_controls:                     sub    $0x4dc,%esp
0xc0ea1526 snd_trident_mixer:                            sub    $0x4c0,%esp
0xc0106307 inflate_fixed:                                sub    $0x4ac,%esp
0xc01080b7 inflate_fixed:                                sub    $0x4ac,%esp
0xc093fff1 ide_config:                                   sub    $0x4a8,%esp
0xc05e404c parport_config:                               sub    $0x490,%esp
0xc0c5ba23 ixj_config:                                   sub    $0x484,%esp
0xc1085e3e gss_pipe_downcall:                            sub    $0x44c,%esp
0xc0b9c9a0 isd200_action:                                sub    $0x444,%esp
0xc03bc848 ciGetLeafPrefixKey:                           sub    $0x428,%esp
0xc04637c3 befs_error:                                   sub    $0x418,%esp
0xc0463833 befs_warning:                                 sub    $0x418,%esp
0xc04638a3 befs_debug:                                   sub    $0x418,%esp
0xc07d27c6 wv_hw_reset:                                  sub    $0x418,%esp
0xc16c9365 root_nfs_name:                                sub    $0x414,%esp
0xc0c7f2d2 bt3c_config:                                  sub    $0x410,%esp
0xc0c833e2 btuart_config:                                sub    $0x410,%esp
0xc0c7d89f dtl1_config:                                  sub    $0x408,%esp
0xc0c816b6 bluecard_config:                              sub    $0x408,%esp

Jörn

-- 
If you're willing to restrict the flexibility of your approach,
you can almost always do something better.
-- John Carmack