Notes: I use gcc 2.6.3 et. al. from Slackware 2.3.0. Non-ELF at this time.
I use the new modules versioning and quite a few modules.
In conclusion, let me just say `thank you!' to everyone who's put in any
work at all on these kernels, especially Linus, and I suggest you don't take
my `tone-of-typer' terribly seriously, because I had a long day yesterday.
:) lilo
First of all, a problem compiling net/ipv4/igmp.c. I was given the
following patch, which seems to do the job:
--- linux/net/ipv4/igmp.c Fri Sep 1 15:37:27 1995
+++ linux/net/ipv4/igmp.c.NEW Fri Sep 1 15:37:27 1995
@@ -106,7 +106,7 @@
}
ih=(struct igmphdr *)skb_put(skb,sizeof(struct igmphdr));
ih- type=IGMP_HOST_MEMBERSHIP_REPORT;
- ih- code=0;
+ ih- unused=0;
ih- csum=0;
ih- group=address;
ih- csum=ip_compute_csum((void *)ih,sizeof(struct igmphdr)); /* Checksum fill */
Next, I discovered that the arcnet stuff (again) doesn't seem to compile as
a module. I don't use it, probably want to and will want it as a module,
but not soon. I'd suggest that someone look at this, and if you need my
messages I'll run another compile. For now I just removed it from
modulization:
--- linux/drivers/net/Makefile Fri Sep 1 21:04:21 1995
+++ linux/drivers/net/Makefile.NEW Fri Sep 1 21:04:21 1995
@@ -245,8 +245,6 @@
ifdef CONFIG_ARCNET
L_OBJS += arcnet.o
-else
-M_OBJS += arcnet.o
endif
ifdef CONFIG_PI
Here's a masquerade patch from Pauline Middelink that still hasn't made it
in. Is there some reason? Please let me know. Masquerade seems to run
fine with it, I haven't tried lately without. :)
--- linux/net/ipv4/ip.c Fri Sep 1 06:31:56 1995
+++ linux/net/ipv4/ip.c.NEW Fri Sep 1 11:56:37 1995
@@ -1819,7 +1819,7 @@
iph- tot_len = ntohs(skb- len-(((unsigned char *)iph)-skb- data));
#ifdef CONFIG_IP_FIREWALL
- if(ip_fw_chk(iph, dev, ip_fw_blk_chain, ip_fw_blk_policy, 0) != 1)
+ if(ip_fw_chk(iph, dev, ip_fw_blk_chain, ip_fw_blk_policy, 0) < 1)
/* just don't send this packet */
return;
#endif
In the `Unnecessary Annoyance' category, someone tell the author of the
latest mcd changes not to kill my modules compile if I haven't looked at his
configuration file. I don't really care all that much that particular CD
driver, and it nailed me twice. :)
--- linux/include/linux/mcd.h Fri Sep 1 17:23:53 1995
+++ linux/include/linux/mcd.h.NEW Fri Sep 1 17:23:53 1995
@@ -29,7 +29,7 @@
/* *** make the following line uncommented, if you're sure,
* *** all configuration is done */
-/* #define I_WAS_HERE */
+#define I_WAS_HERE
/* Increase this if you get lots of timeouts */
#define MCD_STATUS_DELAY 200
--- linux/include/linux/mcdx.h Fri Sep 1 17:38:15 1995
+++ linux/include/linux/mcdx.h.NEW Fri Sep 1 17:38:15 1995
@@ -74,7 +74,7 @@
/* *** make the following line uncommented, if you're sure,
* *** all configuration is done */
-/* #define I_WAS_HERE */
+#define I_WAS_HERE
/* The name of the device */
#define MCDX "mcdx"
And, finally, there's that lovely SLIP problem (possibly something about the
`slip_proto' device?) that results in my not being able to bring up my
system if I have SLIP compiled into the kernel. Someone else has provided
documentation on it in the last few dekamails so I will simply point out
that, if we're going to make SLIP (and PPP) modules, we need to change the
config to give people choices about what those modules look like. I'd
suggest the following change for now, and please let's not get into another
`we need a better kernel configuration database' discussion:
--- linux/arch/i386/config.in Sat Sep 2 10:29:25 1995
+++ linux/arch/i386/config.in.NEW Sat Sep 2 10:29:25 1995
@@ -138,10 +138,14 @@
bool 'SLIP (serial line) support' CONFIG_SLIP n
if [ "$CONFIG_SLIP" = "y" ]; then
bool ' CSLIP compressed headers' CONFIG_SLIP_COMPRESSED y
+else
+ bool 'Compressed headers on SLIP module' CONFIG_SLIP_COMPRESSED y
fi
bool 'PPP (point-to-point) support' CONFIG_PPP n
if [ "$CONFIG_PPP" = "y" ]; then
bool ' 16 channels instead of 4' CONFIG_PPP_LOTS n
+else
+ bool '16 channels instead of 4 on PPP module' CONFIG_PPP_LOTS n
fi
if [ "$CONFIG_AX25" = "y" ]; then
bool 'Z8530 SCC kiss emulation driver for AX.25' CONFIG_SCC y