Date: Mon, 7 Sep 98 19:04 BST From: alan@lxorguk.ukuu.org.uk (Alan Cox) To: linux-kernel@vger.rutgers.edu Subject: Linux 2.0.36pre8 (release candidate one aka 'no chance') Is now on ftp://ftp.linux.org.uk/pub/linux/alan/... The preliminary 2.0.36 release notes are included below and maintained as http://roadrunner.swansea.linux.org.uk/relnotes.36.html Alan ------------------------------------------------------------------ Release Notes For Linux 2.0.36 _________________________________________________________________ Bug Fixes Large Memory Linux 2.0.36 uses the newer bios calls to automatically size memory above 64Mbytes, where supported by the BIOS. Readv/Writev Processes issuing a readv or writev to a device that did not have the relevant read/write operation could cause an Oops and potentially a crash. UDMA Drives Report Bad CRC (cable errors) as a cable error. Handling of such errors was correct. pipe syscall error The pipe call could return ENFILE not -1 and errno= -ENFILE. Fixed. AIC7xxx The 2.0.36 kernel contains the 5.0.20 release of the AIC7xxx driver. This should cure most of the remaining problems with the older chipsets. Users with the latest AIC7xxx devices will have to wait for the next driver version to become available or try the beta test driver. Procfs permissions fix A process with root file rights can now read all /proc files. Daylight Savings in SMB The SMB file system honours daylight savings time. EATA SCSI/Ultrastor 14f/34f Reverse scan order support and configurable extended geometry. Increased the busy timeout. Build fixes Use relative paths for sound, remove trampoline.hex on a make clean, support the newer versioned symbols as part of a multi object file module. Oops handling A small fencepost error in Oops handling on syscall return has been cured. TLan 1.0 The Thunderlan driver has been updated to the 1.0 release. Delay loops The delay loop code has been modified to eliminate most of the remaining cache/branch prediction and other variants to its performance. 5.25" floppy An incorrect floppy table entry has been altered, and a potential crash on unload fixed. Iomega ZIP driver Handle 23.D firmware funnies. Cyclades Serial Upgraded driver from vendor, with assorted bugs fixed. Printer Driver The correction in the printer handling upset the Epson Stylus 800. The driver know has a LPSTRICT option that can be set for printers that need absolute strict NBUSY handling. Beeper gets stuck The case where the beeper decides to beep forever has been cured. 3c509 ethernet Upgraded to v1.16. Fixes ID port clash with sound cards, waiting for discard messages and recovers faster from transmit errors. 3c59x driver Updated 3c59x ethernet driver. This should cure the skb_push panics some people saw with 3c59x/3c90x drivers under load. Lance driver Module unload bugs in the lance driver have been fixed. Plip driver Messages without severity levels now have appropriate levels. AHA1542 Allow the use of DMA 0 on newer motherboards. SHM swap off A case where the machine might crash when turning swapping off has been fixed. MMAP security Linux did not allow a writable mmap of an append only file. It did however allow a readonly mmap of such a file then an mprotect. Fixed. NAKAMICH MJ-5.16S This SCSI device has been added to the blacklist. Single lun blacklisting A bug in the single lun blacklisting has been fixed. SCSI CAM division by zero A corrupt geometry could cause the kernel to divide by zero and crash. Sound blaster Don't report a DMA channel 0 for the MPU Sequencer If there are no synths or midi devices do not lock the sequencer busy forever. Autofs Fix a small glitch in the directory hash. ISOfs Fix a fencepost error in the iso fs size checks. Vfat fs Disallow periods at the end of names. NFS file system Support FIFO's over NFS. Handle kill of nfsiod for module unload. NFS root IFF_MULTICAST could be incorrectly not set. Non modular soundmodem/baycom These devices only worked as modules. They now work compiled in. Memory leak in networking A very obscure leak in the networking code has been fixed. TCP select TCP select for urgent data now has correct semantics. Previously it could do the wrong thing. TCP SIGIO SIGIO on an incoming connection is now correctly issued on the completion of the three way handshake. Enhancements ISDN The Isdn4linux layer is signifcantly upgraded. The new driver set adds support for the Teles 16.3c, Teles PCI, Teles S0Box, Creatix S0Box, Compaq ISDN S0 ISA, ELSA Quickstep 1000PCI, Elsa Quickstep 3000, Elsa Quickstep 3000PCI, Eicon.Diehl Diva 2.0 ISA/PCI (not Pro), Eicon.Diehl Diva Piccola, AsusCom ISDNLink 128K, Dynalink IS64PH, HFC-2BS0 based cards, Sedlbauer Speed Card(Speed win, teledat 100), Sedlbauer Speed Star PCMCIA, USR Sportster Internal TA, ITH MIC 16 ISA, Traverse NETjet PCI, Niccy PnP/PCI Shaper The shaper device provides a simple traffic limiting driver for Linux 2.0.x. For full traffic shaping watch for Linux 2.2 coming soon.. Tulip The tulip driver has been upgraded to 0.89H which should also support the clone PNIC and MXIC tulip devices. Extended CPUID/Chip identification The Cyrix/AMD extended CPUID mode is supported. Cyrix processors are identified even when CPUID is not available. The Intel Celeron Mendicino is recognized. K6's with the random oops bug are normally now detected. The K6 cpu deadlock problem isn't detected as we know no way to check for it except to try it. Multi-Tech driver Driver for multi-tech 4/8 port modem and serial cards.(Experimental). EtherExpress The Compaq LTE should now be recognized (experimental) Intel 440GX The 440GX chipset is known to the PCI data tables. Nvidia NVidia/SGS Thomson is known to the PCI data tables. SCSI medium changers These are no long reported as unknown device types. ROSE networking The ROSE network layer has been updated. Other Bigger system call table To support add ons that use syscalls in the new range. Hooks for dumping Kernel hooks for optional threaded core dump module. - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/faq.html