2.4.23pre6aa2

URL:

	http://www.us.kernel.org/pub/linux/kernel/people/andrea/kernels/v2.4/2.4.23pre6aa2.gz
	http://www.us.kernel.org/pub/linux/kernel/people/andrea/kernels/v2.4/2.4.23pre6aa2/

Pending things for next releases:

o libata for SATA
o UML update

Changelog diff between 2.4.23pre6aa1 and 2.4.23pre6aa2:

Only in 2.4.23pre6aa2: 00_csum-trail-1

Fixup overflow in p6 checksum routing that crashes sendfile, from Andi Kleen. IIRC I found and fixed (though fixed w/o fixmap) this same bug too in 1999, but at that time nobody listened about my claim that it would eventually crash machines due alignment issues.

Only in 2.4.23pre6aa1: 00_extraversion-30 Only in 2.4.23pre6aa2: 00_extraversion-31

Rediffed.

Only in 2.4.23pre6aa2: 70_qsort-1 Only in 2.4.23pre6aa1: 71_qsort-1 Only in 2.4.23pre6aa2: 71_xfs-qsort-1

Allow xfs to compile.

Only in 2.4.23pre6aa1: 91_zone_start_pfn-7 Only in 2.4.23pre6aa2: 91_zone_start_pfn-8

Fix a NUMA bug introduced by this patch that was crashing my opteron desktop. From Andi Kleen.

Only in 2.4.23pre6aa1: 9999_zzz-dynamic-hz-1 Only in 2.4.23pre6aa2: 9999_zzz-dynamic-hz-2

Fixup two miscompiles reported by Eyal Lebedinsky.

Andrea - If you prefer relying on open source software, check these links:

	    rsync.kernel.org::pub/scm/linux/kernel/bkcvs/linux-2.[45]/
	    http://www.cobite.com/cvsps/

This is most unusual as -aa patches usually apply clean, but I am encountering a number of build problems.

"PCMCIA SCSI adapter support" build is broken. I deselected the whole section.

I now have this failure:

gcc -D__KERNEL__ -I/data2/usr/local/src/linux-2.4-pre-aa/include -Wall -Wstrict-
prototypes -Wno-trigraphs -O2 -fno-strict-aliasing -fno-common -fomit-frame-poi
nter -pipe -msoft-float -mpreferred-stack-boundary=2 -march=i686 -malign-functio
ns=4 -DMODULE -DMODVERSIONS -include /data2/usr/local/src/linux-2.4-pre-aa/inclu de/linux/modversions.h -nostdinc -iwithprefix include -DKBUILD_BASENAME=ip_vs_c
onn -c -o ip_vs_conn.o ip_vs_conn.c
ip_vs_conn.c: In function `ip_vs_tunnel_xmit':
ip_vs_conn.c:895: too many arguments to function `ip_select_ident_Rd603b6c5'
make[2]: *** [ip_vs_conn.o] Error 1
make[2]: Leaving directory `/data2/usr/local/src/linux-2.4-pre-aa/net/ipv4/ipvs'

Seems that ip_select_ident() only wants the first argument.

Next I get

gcc -D__KERNEL__ -I/data2/usr/local/src/linux-2.4-pre-aa/include -Wall -Wstrict-
prototypes -Wno-trigraphs -O2 -fno-strict-aliasing -fno-common -fomit-frame-poi
nter -pipe -msoft-float -mpreferred-stack-boundary=2 -march=i686 -malign-functio
ns=4 -DMODULE -DMODVERSIONS -include /data2/usr/local/src/linux-2.4-pre-aa/inclu de/linux/modversions.h -nostdinc -iwithprefix include -DKBUILD_BASENAME=addrcon
f -c -o addrconf.o addrconf.c
addrconf.c: In function `ipv6_store_devconf':
addrconf.c:1996: structure has no member named rtr_solicit_interval' addrconf.c:1997: structure has no member named rtr_solicit_delay' make[2]: *** [addrconf.o] Error 1
make[2]: Leaving directory `/data2/usr/local/src/linux-2.4-pre-aa/net/ipv6'

I think that these members were renamed, my patch is just a guess.

Next
gcc -D__KERNEL__ -I/data2/usr/local/src/linux-2.4-pre-aa/include -Wall -Wstrict-
prototypes -Wno-trigraphs -O2 -fno-strict-aliasing -fno-common -fomit-frame-poi
nter -pipe -msoft-float -mpreferred-stack-boundary=2 -march=i686 -malign-functio
ns=4 -DMODULE -DMODVERSIONS -include /data2/usr/local/src/linux-2.4-pre-aa/inclu de/linux/modversions.h -nostdinc -iwithprefix include -DKBUILD_BASENAME=protoco
l -c -o protocol.o protocol.c
protocol.c: In function `sctp_v4_create_accept_sk':
protocol.c:537: structure has no member named `id'
make[2]: *** [protocol.o] Error 1
make[2]: Leaving directory `/data2/usr/local/src/linux-2.4-pre-aa/net/sctp'

Member .id was removed, so my patch removes the offending line (a guess).

Finally I get
if [ -r System.map ]; then /sbin/depmod -ae -F System.map 2.4.23-pre6-aa2; fi
depmod: /lib/modules/2.4.23-pre6-aa2/ksyms is not an ELF file
depmod: /lib/modules/2.4.23-pre6-aa2/soundconf is not an ELF file
depmod: *** Unresolved symbols in /lib/modules/2.4.23-pre6-aa2/kernel/drivers/vi deo/sis/sisfb.o

depmod:         __floatsidf
depmod:         __divdf3
depmod:         __fixunsdfsi
depmod:         __muldf3
depmod:         __adddf3

And building i2c-2.7.0 (which I need for sensors) is failing.

depmod: *** Unresolved symbols in
/lib/modules/2.4.23-pre6-aa2/kernel/drivers/ie
ee1394/pcilynx.o

depmod:         i2c_bit_add_bus_Rca543f36
depmod:         i2c_transfer_R1dea91d1
depmod:         i2c_bit_del_bus_Rdf920b11
depmod: *** Unresolved symbols in
/lib/modules/2.4.23-pre6-aa2/kernel/drivers/me
dia/video/bttv.o
depmod:         i2c_bit_add_bus_Rca543f36
depmod:         i2c_master_recv_R67b29cc4
depmod:         i2c_bit_del_bus_Rdf920b11
depmod:         i2c_master_send_Rb692cb0e
depmod: *** Unresolved symbols in
/lib/modules/2.4.23-pre6-aa2/kernel/drivers/me
dia/video/msp3400.o
depmod:         i2c_probe_R4e2acbec
depmod:         i2c_add_driver_Racf22304
depmod:         i2c_transfer_R1dea91d1
depmod:         i2c_attach_client_Ra861362d
depmod:         i2c_master_send_Rb692cb0e
depmod:         i2c_detach_client_R0cfb40b4
depmod:         i2c_del_driver_R57837012

.... more following, all in kernel/drivers/media/video/* ...

--
Eyal Lebedinsky ([email protected]) <http://samba.org/eyal/>

@@ -892,7 +892,7 @@
 	iph->saddr		=	rt->rt_src;
 	iph->ttl		=	old_iph->ttl;
 	iph->tot_len		=	htons(skb->len);
-	ip_select_ident(iph, &rt->u.dst, NULL);
+	ip_select_ident(iph);
 	ip_send_check(iph);
 
 	skb->ip_summed = CHECKSUM_NONE;

@@ -1993,8 +1993,8 @@
 	array[DEVCONF_AUTOCONF] = cnf->autoconf;
 	array[DEVCONF_DAD_TRANSMITS] = cnf->dad_transmits;
 	array[DEVCONF_RTR_SOLICITS] = cnf->rtr_solicits;
-	array[DEVCONF_RTR_SOLICIT_INTERVAL] = cnf->rtr_solicit_interval;
-	array[DEVCONF_RTR_SOLICIT_DELAY] = cnf->rtr_solicit_delay;
+	array[DEVCONF_RTR_SOLICIT_INTERVAL] = cnf->__rtr_solicit_interval;
+	array[DEVCONF_RTR_SOLICIT_DELAY] = cnf->__rtr_solicit_delay;
 #ifdef CONFIG_IPV6_PRIVACY
 	array[DEVCONF_USE_TEMPADDR] = cnf->use_tempaddr;
 	array[DEVCONF_TEMP_VALID_LFT] = cnf->temp_valid_lft;

@@ -534,7 +534,6 @@
 	newsk->dport = htons(asoc->peer.port);
 	newsk->daddr = asoc->peer.primary_addr.v4.sin_addr.s_addr;
 	newinet->pmtudisc = inet->pmtudisc;
-      	newinet->id = 0;
 
 	newinet->ttl = sysctl_ip_default_ttl;
 	newinet->mc_loop = 1;

can you double check you can reproduce it in 2.4.23pre6 vanilla? It doesn't seem introduced by my patches.

this isn't new, we simplified the ip_select_ident a long time ago. Not sure it triggers only now. Your fix is correct thanks.

this one is related to dynamic-hz so you're right it's a new thing. But careful, your fix is wrong. The rename was done for a reason, that is exactly to catch those at compile time. This will fix it right:

@@ -1993,8 +1993,8 @@ static void inline ipv6_store_devconf(st
 	array[DEVCONF_AUTOCONF] = cnf->autoconf;
 	array[DEVCONF_DAD_TRANSMITS] = cnf->dad_transmits;
 	array[DEVCONF_RTR_SOLICITS] = cnf->rtr_solicits;
-	array[DEVCONF_RTR_SOLICIT_INTERVAL] = cnf->rtr_solicit_interval;
-	array[DEVCONF_RTR_SOLICIT_DELAY] = cnf->rtr_solicit_delay;
+	array[DEVCONF_RTR_SOLICIT_INTERVAL] = user_to_kernel_hz_overflow(cnf->__rtr_solicit_interval);
+	array[DEVCONF_RTR_SOLICIT_DELAY] = user_to_kernel_hz_overflow(cnf->__rtr_solicit_delay);
 #ifdef CONFIG_IPV6_PRIVACY
 	array[DEVCONF_USE_TEMPADDR] = cnf->use_tempaddr;
 	array[DEVCONF_TEMP_VALID_LFT] = cnf->temp_valid_lft;

Well SCPT is a new feature, introduced in 2.4.23pre, this had nothing to do with my tree, but thanks for the fix, merged.

those drivers are buggy, and had always been buggy, it's just that you couldn't notice it before. I added an option in my new tree to catch those longstanding bugs. They *must* not compile. Driver authors can audit their 2.4 drivers by grabbing my tree or by applying this patch alone:

	http://www.us.kernel.org/pub/linux/kernel/people/andrea/kernels/v2.4/2.4.23pre6aa2/9999900_soft-float-1

see drivers/video/sis/init.c:

      divider = (sr2b & 0x80) ? 2.0 : 1.0;
      postscalar = (sr2c & 0x80) ?
              ( (((sr2c >> 5) & 0x03) == 0x02) ? 6.0 : 8.0) : (((sr2c >> 5) & 0x03) + 1.0);
      num = (sr2b & 0x7f) + 1.0;
      denum = (sr2c & 0x1f) + 1.0;

there would be a slight chance to find false positives too, but those aren't false positives, there's no fpu save at all there.

Ironically all the float numbers seems to end with .0, I fixed it a bit but I've no idea if I catched all them, so Eyal, you can try to apply this patch and follow the code to see if you can spot more of these longstanding bugs (those could corrupt the fpu state and segfault userspace at the very least):

@@ -3940,11 +3940,11 @@ SiSBuildBuiltInModeList(ScrnInfoPtr pScr
       sr2b = pSiS->SiS_Pr->SiS_VCLKData[vclkindex].SR2B;
       sr2c = pSiS->SiS_Pr->SiS_VCLKData[vclkindex].SR2C;
 
-      divider = (sr2b & 0x80) ? 2.0 : 1.0;
+      divider = (sr2b & 0x80) ? 2 : 1;
       postscalar = (sr2c & 0x80) ?
-              ( (((sr2c >> 5) & 0x03) == 0x02) ? 6.0 : 8.0) : (((sr2c >> 5) & 0x03) + 1.0);
-      num = (sr2b & 0x7f) + 1.0;
-      denum = (sr2c & 0x1f) + 1.0;
+              ( (((sr2c >> 5) & 0x03) == 0x02) ? 6 : 8) : (((sr2c >> 5) & 0x03) + 1);
+      num = (sr2b & 0x7f) + 1;
+      denum = (sr2c & 0x1f) + 1;
 
 #ifdef TWDEBUG
       xf86DrvMsg(0, X_INFO, "------------\n");

this looks like if you didn't compile the needed i2c (or maybe it was due the lack of a make dep first), the above modules (pcilynx bttv msp3400) looks innocent.

Thanks for the help in tracking and fixing those bits!

Andrea - If you prefer relying on open source software, check these links:

	    rsync.kernel.org::pub/scm/linux/kernel/bkcvs/linux-2.[45]/
	    http://www.cobite.com/cvsps/

I am reasonably sure that vanilla -pre6 built without any complaint.

I have a script that goes into the i2c-2.7.0 tree and builds it. This worked just fine with vanilla -pre6 (and older ones for a long time now) but fails with -pre6-aa2.

Nevertheless, I am now building vanilla -pre6 just to confirm the above claims... built just fine. Retry -aa2... failed.

Here is the diff in .config between my vanilla -pre6 and -aa2

@@ -44,8 +44,7 @@
CONFIG_X86_XADD=y
CONFIG_X86_BSWAP=y
CONFIG_X86_POPAD_OK=y
-# CONFIG_RWSEM_GENERIC_SPINLOCK is not set
-CONFIG_RWSEM_XCHGADD_ALGORITHM=y
+CONFIG_X86_CMPXCHG8=y
CONFIG_X86_L1_CACHE_SHIFT=6
CONFIG_X86_HAS_TSC=y
CONFIG_X86_GOOD_APIC=y
@@ -56,6 +55,7 @@
CONFIG_X86_MCE=y
CONFIG_TOSHIBA=m
CONFIG_I8K=m
+CONFIG_THINKPAD=m
CONFIG_MICROCODE=m
CONFIG_X86_MSR=m
CONFIG_X86_CPUID=m
@@ -63,6 +63,11 @@
# CONFIG_HIGHMEM4G is not set
# CONFIG_HIGHMEM64G is not set
# CONFIG_HIGHMEM is not set
+CONFIG_FORCE_MAX_ZONEORDER=11
+CONFIG_1GB=y
+# CONFIG_2GB is not set
+# CONFIG_3GB is not set
+# CONFIG_05GB is not set
# CONFIG_MATH_EMULATION is not set
CONFIG_MTRR=y
# CONFIG_SMP is not set
@@ -105,6 +110,8 @@
CONFIG_SYSVIPC=y
# CONFIG_BSD_PROCESS_ACCT is not set
CONFIG_SYSCTL=y
+CONFIG_MAX_USER_RT_PRIO=100
+CONFIG_MAX_RT_PRIO=0
CONFIG_KCORE_ELF=y
# CONFIG_KCORE_AOUT is not set
CONFIG_BINFMT_AOUT=m
@@ -316,6 +323,7 @@
CONFIG_MD_RAID5=m
CONFIG_MD_MULTIPATH=m
CONFIG_BLK_DEV_LVM=m
+CONFIG_BLK_DEV_DM=m

#
# Networking options
@@ -328,6 +336,10 @@
CONFIG_FILTER=y
CONFIG_UNIX=m
CONFIG_INET=y
+CONFIG_TUX=m
+CONFIG_TUX_EXTCGI=y
+# CONFIG_TUX_EXTENDED_LOG is not set
+# CONFIG_TUX_DEBUG is not set
# CONFIG_IP_MULTICAST is not set
# CONFIG_IP_ADVANCED_ROUTER is not set
# CONFIG_IP_PNP is not set
@@ -715,6 +727,10 @@
CONFIG_SCSI_QLOGIC_FC=m
# CONFIG_SCSI_QLOGIC_FC_FIRMWARE is not set
CONFIG_SCSI_QLOGIC_1280=m
+CONFIG_SCSI_QLOGIC_QLA2XXX=y
+CONFIG_SCSI_QLOGIC_QLA2XXX_QLA2100=m
+CONFIG_SCSI_QLOGIC_QLA2XXX_QLA2200=m
+CONFIG_SCSI_QLOGIC_QLA2XXX_QLA2300=m
CONFIG_SCSI_SEAGATE=m
CONFIG_SCSI_SIM710=m
CONFIG_SCSI_SYM53C416=m
@@ -731,11 +747,7 @@
#
# PCMCIA SCSI adapter support
#
-CONFIG_SCSI_PCMCIA=y
-CONFIG_PCMCIA_AHA152X=m
-CONFIG_PCMCIA_FDOMAIN=m
-CONFIG_PCMCIA_NINJA_SCSI=m
-CONFIG_PCMCIA_QLOGIC=m
+# CONFIG_SCSI_PCMCIA is not set

#
# Fusion MPT device support
@@ -909,6 +921,7 @@
CONFIG_FDDI=y
CONFIG_DEFXX=m
CONFIG_SKFP=m
+CONFIG_NETCONSOLE=m
CONFIG_HIPPI=y
CONFIG_ROADRUNNER=m
# CONFIG_ROADRUNNER_LARGE_RINGS is not set
@@ -1479,6 +1492,7 @@
# CONFIG_AGP_SWORKS is not set
# CONFIG_AGP_NVIDIA is not set
# CONFIG_AGP_ATI is not set
+CONFIG_ECC=m

#
# Direct Rendering Manager (XFree86 DRI support)
@@ -1619,6 +1633,13 @@
# CONFIG_UDF_RW is not set
CONFIG_UFS_FS=m
# CONFIG_UFS_FS_WRITE is not set
+CONFIG_XFS_FS=m
+# CONFIG_XFS_POSIX_ACL is not set
+# CONFIG_XFS_RT is not set
+# CONFIG_XFS_QUOTA is not set
+# CONFIG_XFS_DMAPI is not set
+# CONFIG_XFS_DEBUG is not set
+# CONFIG_PAGEBUF_DEBUG is not set

#
# Network File Systems
@@ -2061,3 +2082,4 @@
CONFIG_ZLIB_INFLATE=m
CONFIG_ZLIB_DEFLATE=m
CONFIG_FW_LOADER=m
+CONFIG_QSORT=m

--
Eyal Lebedinsky ([email protected]) <http://samba.org/eyal/>

[trimmed]

OK, carefully inspecting the logs explains the above. -aa2 breaks the build
of i2c-2.7.0, but my script already removed the original i2c modules by then. The failure is just another static HZ initializer situation.

I think that we need an option to revert HZ to a constant for people that do not want to fight with this change for now (I am sure there are other drivers in the wild that will take a while to catch up with this change even after it makes mainline).

gcc -I/usr/src/linux/include -O2 -DLM_SENSORS -D__KERNEL__ -DMODULE -fomit-frame
-pointer -DEXPORT_SYMTAB -DMODVERSIONS -include /usr/src/linux/include/linux/mod
versions.h -c kernel/i2c-philips-par.c -o kernel/i2c-philips-par.o kernel/i2c-philips-par.c:163: initializer element is not constant kernel/i2c-philips-par.c:163: (near initialization for `bit_lp_data.timeout')
kernel/i2c-philips-par.c:172: initializer element is not constant kernel/i2c-philips-par.c:172: (near initialization for `bit_lp_data2.timeout')

--
Eyal Lebedinsky ([email protected]) <http://samba.org/eyal/>