Marcelo Tosatti [interview] released 2.4.28-pre4 with few changes since -pre3 a month ago [story], "it contains a number of driver updates (pcnet, e1000, gdth, prism54), a network update from David, few more gcc3.4 warning fixes." He noted that no more new features are planned for the 2.4 stable kernel, "from now on [I] can change only what is necessary and let the 2.4 tree [rest] in peace :)"
From: Marcelo Tosatti [email blocked] To: linux-kernel Subject: Linux 2.4.28-pre4 Date: Fri, 8 Oct 2004 08:21:35 -0300 Hi, Here goes 2.4.28-pre4... It contains a number of driver updates (pcnet, e1000, gdth, prism54), a network update from David, few more gcc3.4 warning fixes. I'm happy that the number of updates is small, -pre3 has been released more than one month ago. >From now on can now change only what is necessary and let the 2.4 tree in peace :) Summary of changes from v2.4.28-pre3 to v2.4.28-pre4 ============================================ <ajgrothe:yahoo.com>: o [CRYPTO]: Whirlpool algorithm updates o [CRYPTO]: Add missing tcrypt part of whirlpool updates <ananth:broadcom.com>: o [libata sata_svw] race condition fix, new device support <joshk:triplehelix.org>: o radeonfb: Fix module unload and red/blue typo o hotplug: Don't build cpqphp_proc.o if !PROC_FS <lesanti:sinectis.com.ar>: o fix dcache nr_dentry race <martin.wilck:fujitsu-siemens.com>: o [TG3]: Fix pause handling, we had duplicate flags for the same thing <michael.waychison:sun.com>: o [TG3]: Fix thinko in 5704 fibre hw autoneg code <peter:pantasys.com>: o [IPCONFIG]: Verify DHCPACK packets o [IPV4]: Fix DHCPACK checking in ipconfig.c <tkooda-patch-kernel:devsec.org>: o [CRYPTO]: xtea_encrypt() should use XTEA_DELTA instead of TEA_DELTA <vda:port.imtp.ilyichevsk.odessa.ua>: o trivial patch for 2.4: always inline __constant_* Achim Leubner: o gdth update David S. Miller: o [NET]: Kill SCM_CONNECT, never used and unreferenced o [TCP]: Just silently ignore ICMP Source Quench messages o [TG3]: Recognize all onboard Sun variants, not just 5704 o [TG3]: Update driver version and reldate o [CRYPTO]: Zero out tfm before freeing in crypto_free_tfm() o [SPARC64]: Do not log streaming byte hole errors o [PKT_SCHED]: sch_netem.c needs linux/init.h o [SPARC64]: Disable SBH interrupt properly David Woodhouse: o [NET]: In compat syscall handling, check socket option types correctly Don Fry: o pcnet32: discard oversize rx packets o pcnet32: recover after rx hang o pcnet32: cleanup IRQ limitation o pcnet32: Add HomePNA parameter for 79C978 o pcnet32: correctly program bcr32 Doug Ledford: o RAID1 error handling locking fix Ganesh Venkatesan: o e1000 - ethtool support cleanup o e1000 - Enable TSO o e1000 - Replace kmalloc with vmalloc for data structures not shared with h/w o e1000 - TSO context descriptor setup fixes (in preparation for IPv6 TSO) o e1000 - Fix to prevent infinite loop trying to re-establish link while actively communicating o e1000 - Condition that determines when to quit polling mode includes work done in Tx path o e1000 - Shutdown PHY while bringing the interface down (if WoL not enabled) o e1000 - add likely/unlikely to assist branch prediction, other cleanups o e1000 - more DPRINTK messages o e1000 - suspend/resume fix from [email blocked] o e1000 - white space corrections o e1000 - remove support for advanced TCO features o e1000 - Fix MODULE_PARM, module_param and module_param_array usage o e1000 - Fix VLAN filter setup errors (while running on PPC) o e1000 - Polarity reversal workaround for 10F/10H links o e1000 - white space corrections, other cleanups o e1000 update - reset default ITR value to 8000 Geert Uytterhoeven: o m68k MM off-by-one o Atari ST-RAM setup o Amiga frame buffer: kill obsolete DMI Resolver code o fbdev monochrome lines Herbert Xu: o Backport Via IRQ mask fix Hideaki Yoshifuji: o [IPV6] Fix routing header handling o [IPV6] Fix skb allocation size for RST and ACK o [IPV6]: Missing ip_rt_put() in SIT error path Jack Hammer: o broken ips update Jean Delvare: o Update Documentation/i2c/writing-clients Jeff Garzik: o [TG3]: Kill all on-chip send BD support code o linux/compiler.h: dummy __iomem macro (an sparse annotation) o [libata] resync with 2.6.x o [libata] remove distinction between MMIO/PIO helper functions o [libata] consolidate legacy/native mode init code into helpers o [libata] minor comment updates, preparing for iomap merge Jens Axboe: o irq safe gendisk_lock Linus Torvalds: o libata: initial PCI memory annotations Marcelo Tosatti: o Cset exclude: [email blocked]|ChangeSet|20040928105422|00490 o Mike Miller: cciss typo fix o Changed EXTRAVERSION to -pre4 Margit Schubert-While: o prism54 Code cleanup o prism54 remove module params o prism54 add WE17 support o prism54 initial WPA support o prism54 fix wpa_supplicant frequency parsing o prism54 remove TRACE o prism54 Bug in timeout scheduling o prism54 print firmware version o prism54 bug initialization/mgt_commit Maximilian Attems: o menuconfig fix crash due to infinite recursion Mikael Pettersson: o 53c700 scsi driver gcc-3.4 fixes o pcmcia mem_op.h gcc-3.4 fixes o ATM drivers gcc-3.4 fixes o IBM PCI hotplug controller driver gcc-3.4 fixes o ISDN drivers gcc-3.4 fixes o MTD drivers gcc-3.4 fixes o RIVA driver gcc-3.4 fix o E100 driver gcc-3.4 fixes o PPC32 PReP residual data gcc-3.4 fix o matrox framebuffer driver gcc-3.4 fix Pete Zaitcev: o USB drivers gcc-3.4 fixes Stephen Hemminger: o [TCP]: Store congestion algorithm per socket o [TCP]: Add vegas style bandwidth info to 2.4.x tcp diag o [TCP]: Backport 2.6.x cleanup of westwood code Thomas Graf: o [PKT_SCHED]: Fix slab corruption in cbq_destroy o [PKT_SCHED] Report qdisc parent to userspace Wensong Zhang: o [IPVS] add the MAINTAINERS entry
Big thanks to Marcelo
I like to say thanks to Marcelo for producing a truly stable
2.4 series of kernel releases. Big cheer to Marcelo!
Now I'm hoping that 2.6 series will settle down and 2.7 can
start. Hopefully 2.7 will have a shorter development period
that 2.5.
sig = 0xda1e;Different dev cycle
Well, appartenly, your wishes won't be fullfiled since the new dev cycle of 2.6 doesn't make the creation of an experiemental necessary since Andrew's tree is used for that purpose...
Re: Different dev cycle
Yes, I know about that, I'm still not convinced of the new development model. But even so they have said any really big updates will result in a 2.7 fork.
sig = 0xda1e;Re:
Well this approach has allowed Linus to fix performance and other regressions much more quickly then before. And because patches actually get accepted and tested by more people and at a faster rate. More developers are encouraged to create patches. In 2.4 I think it took longer to mature because it had such a slow rate of patch application.
With 2.6 you've got a lot of janitor and other patches getting applied much more quickly.
R.I.P.
R.I.P kernel 2.4 and thanks for the smooth* ride :)
(*except for the version that were borked and ate your FS and stuff)
R.I.P.?
Given that there are still people using and even updating 2.0.x kernels (see http://kerneltrap.org/node/view/2279 ) I think saying RIP to 2.4.x is a bit premature... Rather, 2.4.x can finally be called a "stable kernel" :-)