Changes from 1.1.89 to 1.1.90

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 patch90.gz

• Fix the makefile so that if the Aztech CD driver is included, so is the iso9660 fs.
• Be sure to zero out the LDTs after vmalloc'ing them.
• Turn off the trace flag when handling interrupts.
• Always generate a segmentation violation when we handle an i386 general protection fault.
• More floppy boot-time options added to .../drivers/block/README.fd.
• Support two floppy controllers at runtime instead of compile-time.
• Remove the following two changes added in 1.1.88:
• -Handle slow serial devices by waiting extra time after the receiver gets closed.
• -Add support to detect the StarTech 16650 chip. Treat it as a 16450
• - for now, because of its FIFO bugs.
• Another cheapo timeout in the keyboard handler increased by an order of magnitude.
• Change Alpha support for keyboard interrupts.
• 3c501 cleaned up a little.
• Change de600 and de620 cleanup_module() to match the standard.
• PLIP basically rewritten.
• EATA/DMA SCSI host adapter, cleared target_time_out counter while performing a reset, and all external symbols renamed to avoid possible name conflicts. Now version 1.15.00.
• Oops, unregistering the first SCSI device would remove all from the list.
• Don't reset SCSI devices too often (every 10 sec).
• In some systems, using two DMA channels simultaneously causes unpredictable results. Keep a list of SCSI hosts that use DMA, and block out concurrent operation.
• UltraStor 14F/34F now clears target_time_out counter while performing a reset.
• isofs multisession code cleaned up.
• removed call to socki_lookup in NFS inode handler (use the struct entry instead).
• More PCI device IDs added (surely there's a better way to get this information?)
• Beginning of the tweaking of the swap code. Continued changes through 1.1.94. Too many changes for me to grok. Basically, it's different. Deal.
• Before accepting an ICMP redirect, make sure it's not redirecting *our* address.
• If we close a socket from CLOSE_PENDING, be sure to mark it as dead.
• If we're still trying to open a socket, but we retransmitted too many times, be sure to discard the retransmit timer.
• If we couldn't get more memory for a socket, or if we couldn't build the header for an outgoing packet be sure to mark the state as TCP_CLOSE, so that it goes away.

[Index][Next]