Hello kernel hackers. =] I'm running 2.1.32 now on my 6x86 and it
mostly looks very good. It built and booted fine, of course, having
payed close enough attention to `make config` to see that console bootup
is N by default. ;]
My reservations presently are as follows:
1.) There are a lot of unresolved symbols in drivers built as modules.
Check this `depmod -aev` out:
*** Unresolved symbols in module /lib/modules/2.1.32/fs/nfs.o
nlmclnt_proc
rpc_wake_up_task
rpc_shutdown_client
lockd_down
lockd_up
rpcauth_matchcred
rpcstat_unregister
rpc_sleep_on
rpcstat_get_info
rpc_init_task
rpc_release_task
waitqueue_lock
rpc_execute
rpc_wake_up_next
rpciod_up
xprt_destroy
rpcauth_lookupcred
rpciod_down
rpc_allocate
xprt_create_proto
rpc_do_call
rpc_call_setup
rpcstat_register
xdr_encode_string
nfs_debug
rpc_create_client
*** Unresolved symbols in module /lib/modules/2.1.32/misc/lp.o
__global_cli
__global_sti
*** Unresolved symbols in module /lib/modules/2.1.32/misc/netlink.o
waitqueue_lock
__global_cli
__global_save_flags
__global_restore_flags
__global_sti
synchronize_irq
*** Unresolved symbols in module /lib/modules/2.1.32/scsi/aic7xxx.o
__global_save_flags
__global_cli
__global_restore_flags
*** Unresolved symbols in module /lib/modules/2.1.32/scsi/scsi_mod.o
__global_cli
__global_restore_flags
__global_sti
waitqueue_lock
__global_save_flags
local_irq_count
*** Unresolved symbols in module /lib/modules/2.1.32/scsi/sg.o
__global_save_flags
__global_cli
__global_restore_flags
waitqueue_lock
*** Unresolved symbols in module /lib/modules/2.1.32/scsi/sr_mod.o
__global_cli
__global_restore_flags
__global_save_flags
local_irq_count
*** Unresolved symbols in module /lib/modules/2.1.32/block/floppy.o
__global_save_flags
__global_cli
__global_restore_flags
tqueue_lock
__global_sti
*** Unresolved symbols in module /lib/modules/2.1.32/ipv6/ipv6.o
__global_cli
__global_restore_flags
synchronize_irq
__global_save_flags
/lib/modules/2.1.32/fs/binfmt_aout.o
/lib/modules/2.1.32/fs/binfmt_java.o
/lib/modules/2.1.32/fs/fat.o
/lib/modules/2.1.32/fs/isofs.o
/lib/modules/2.1.32/fs/minix.o
/lib/modules/2.1.32/fs/msdos.o
/lib/modules/2.1.32/fs/smbfs.o
/lib/modules/2.1.32/fs/umsdos.o
/lib/modules/2.1.32/fs/vfat.o
/lib/modules/2.1.32/net/bsd_comp.o
/lib/modules/2.1.32/net/dummy.o
/lib/modules/2.1.32/net/tunnel.o
/lib/modules/2.1.32/scsi/scsi_syms.o
/lib/modules/2.1.32/block/ide-cd.o
/lib/modules/2.1.32/block/loop.o
/lib/modules/2.1.32/block/rd.o
/lib/modules/2.1.32/cdrom/cdrom.o
/lib/modules/2.1.32/ipv4/ip_alias.o
/lib/modules/2.1.32/ipv4/ip_masq_ftp.o
/lib/modules/2.1.32/ipv4/ip_masq_irc.o
/lib/modules/2.1.32/ipv4/ip_masq_quake.o
/lib/modules/2.1.32/ipv4/ip_masq_raudio.o
/lib/modules/2.1.32/ipv4/ipip.o
I admit I'm unqualified to solve these problems on my own, or even
complain about them, for that matter. But, for whatever it's worth, I
humbly suggest that some new theory be put into the construction of
modules during compile time that can better catch this problem. We have
really been slammed by this one on almost every 2.1 kernel we've had. It
looks like a lot of the unresolved symbols in the example I posted are the
same.
2.) I'm wondering when (if ever) the Cyrix-enhancing patches are going to
go in. It's been quite a while and the last one posted was very nice. Is
there a political reason for not merging them in?
Thank you very much, and I hope that my lengthy post is more helpful than
annoying. =]
-bp
-- # B. James Phillippe # Network/Sys Admin Terran.ORG # # <bryan@terran.org> # http://w3.terran.org/~bryan # # Finger for PGP key # Linux Convert - circa 1.1.59 # # Systems Administrator, Seattle Software Labs, Inc #