Changes from 1.3.3 to 1.3.4

These changes are archived on ftp.crynwr.com:pub/kchanges and http://www.crynwr.com/kchanges. I cannot answer questions about Linux -- I merely summarize the kernel patches after reading them. I try to make them useful to everyone, but kernel hackers should take them with a grain of salt and read the patches themselves. The patch file is patch-1.3.4.gz

• Added Andrew Tridgell to CREDITS
• Lots of stuff added to alpha port.
• Adaptec AHA274X/284X/294X support added to i386/config.in.
• EtherExpressPro support uncommented from i386/config.in.
• UMSDOS fixed and uncommented from i386/config.in.
• Oops! Bit tests (foo&bar<baz) not parenthesized correctly in binfmt_elf.c and in mips/kernel/traps.c.
• sys_mmap is being obsoleted (renamed to old_mmap and moved to sys_i386.c).
• sys_pipe moved from i386/kernel/process.c to i386/process/sys_i386.c
• Much work on floppy driver:
  • Dynamically allocate the floppy track buffer.
  • Floppy driver modularized.
  • added support for Commodore 1581 3.5" disks.
  • Made more portable (for Alpha floppies).
  • Added lots of sanity checks.
  • Fixed (and cleaned up) the way the timeout was reset.
• Alphas don't have any CMOS parameters.
• IDE driver made more portable for Alpha.
• Avoid dnagling IRQs in the IDE IRQ probe.
• Lots of Sparc progress -- maybe it boots now?
• Queue up to 8 I/O requests for swapfile I/O.
• Oops, fix a problem testing for xd I/O overlapping a 64K boundary.
• Lots of work done renaming get_fs_* to get_user and put_fs_* to put_user. This patch hurt, but it was worth doing.
• lp modularization cleaned up.
• Oops, found a bitmask-anding problem in scc driver.
• Add EEXPRESS_PRO to modules list if it isn't included.
• ARCNET fixed to make exception packets from us receivable on other systems.
• EtherExpress Pro driver added.
• Network Equalizer problem detecting down interfaces.
• Oops, minor problem in ne2 driver where we tried to reset the whole device instead of just resetting the interrupt status register.
• Schneider & Koch G16 Ethernet driver didn't properly catch giant packets.
• Supplying a commandline argument to the ncr53c8xx driver didn't override the device and function properly.
• ncr53c7,8xx driver cleaned up, and a call to mb() added in various places. Old size not being restored correctly due to unparenthesized &.
• Adaptec 274x driver changed into a 274x/284x/294x driver, and renamed internally to aic7xxx.
• Future Domain TMC-36C70 support added.
• SCSI CDROM now refuses to eject a CD-ROM if multiple parties have it open.
• Skip MSDOS filesystem files that have the SYS or HIDDEN attribute (as well as VOLUME, which is already did).
• Fix a security hole in the proc fs (make sure we're who we say we are).
• If a fs has a llseek handler, use it, but only for signed long offsets.
• Only update the file position if it's actually changed.
• Changed the way umsdos_readdir works.
• A few alpha changes.
• Portability macros added to i386.
• Inline i386 code changed in a way I don't understand.
• Lots of work on sparc arch.
• Change the get_free_pages() and get_dma_pages() routines to allocate below a certain limit. For get_free_pages, it's the end of memory. For get_dma_pages, it's the highest address a DMA device can reach (16Mbytes on i386). Also, don't swap out DMA pages.
• Took all the floppy tape drive definitions out (don't need 'em in the kernel).
• Changed a few unsigned long's in sock.h to __u32.
• Found a bug in the shared memory swapper where it wasn't catching the wrong id number.
• Oops, caught some i386-dependent code in module handler. Also made some code more portable by using manifest constants.
• ICMP Source Quench wasn't working right.

[Index][Next]