NEC claims it has the highest energy density of any electrochemical device, as well as a current discharge rate 20 times
higher than lead batteries (20mAh device: 9A in 10 seconds... electric cars,
anyone?)
Hmmm, 9A in 10 sec = 25 mAh, that is more than the
20 mAh battery capacity. There is something wrong here !
Buy the PC without a harddrive, and buy the
harddrive separate. I did that when I bought my laptop which
of course is Micros~1 free. I didn't even have to return any
CD with that legacy OS !!
I did a quick survey on the "Norman
virus scanner, it knows about 18.284 viruses, that is more
than 25% of all programs written according to the "Cato Institute".
That simply has to be wrong, their estimate of 70.000 programs is way
to low.
> 70 dBm is 70 dB (=10000000 times) above 1 mW = 10 W.
> 70 dBW is 70dB above 1 W = 10kW.
Gee, where was I when I wrote that ?
Yes, 70dBm is 10kW, but it only makes the case more clear. By using
VMSK/2 modulation you use a lot of power to conserve bandwidth. This
can be advantageous in some situations, but clearly *not* when you
need high baud rates
C = capacity in bps
B = bandwidth
SNR = signal to noise ratio
Solving for SNR in dB ( = 10*log10(SNR) ) gives:
SNR_db = 10 * log10 (( 2 ^ (C/B)) - 1)
With C = 12.8 Mbit/s and B = 200 kHz you get SNR = 192 dB !
To have a SNR = 192 dB, the signal has to be 1.8 * 10^19 times
stronger than the noise ! A receiver with 200kHz bandwidth will
typically have a noise floor at -120 dBm, so you need more than
70 dBm received signal strength. 70 dBm is 10 W !! And that is
the signal strength at the receiving antenna, so the transmitter
would have to be in the gigawatt range, to reach short distances.
This means that VMSK/2 can be used, but you can't reach 12.8 Mbit/s
without a nuclear powered transmitter. You can get a decent bit rate with VMSK/2
on battery powered equipment, but you have to design for a few kbit/s,
not 12.8 Mbit/s. Nice theory, but....
Does this mean that RH7.0 will be a final snapshot from rawhide, and not a upgraded 7.0beta ? In a few weeks when the bug reports starts ticking in, will rawhide be more stable than 7.0beta, or is rawhide always the bleeding edge unstable redhat where we can crashtest all the new stuff ?
Cheops is a network "swiss army knife". It's "network neighborhood" done right (or gone out of control, depending on your perspective). It seems that the development has slowed down a bit though.
you lie about your ip address (that's called spoofing). A simple traceroute will give the same information if you don't spoof your ip. If you spoof your ip, the server you are connecting to can't send anything back to you since it doesn't know where to send it.
you send massive amounts of data. The itrace message is only sent for every 20000 packet. That is flooding.
Itrace will only give useful information about users who are abusing the flaws in the tcp/ip protocol and at the same time are sending lots of data (like syn flooding), and will not have any consequences for the rest of us.
I like simple protocols/standards with no possibilities for extensions. Look at history:
Kerberos: It has a field which is reserved. The standard was abused by MS, and the extension is now a "trade secret" making "MS-Kerberos" a property protocol.
SMTP mail: Can be extended with X- headers. Used is many incompatible manners by diverse email software.
HTML: A standard which is designed to be easy to extend. Need I say more ?
Keep the protocols simple with no possibilities for extensions, and make sure that implementations don't fail gracefully when abused. If so, all implementations will work together, and MS and other will fail when they apply their "embrace and extend" tactics.
This reminds me of the first radio amateurs. They where assigned all frequencies above 300 kHz (wavelength shorter than 1 km) because the frequency was so high that (it was believed at that time) "the radio waves propagates as light" and thus could not pass any obstacles like trees, buildings and the horizon.....
It could not have been more wrong, and today the radio amateurs are crammed into small segments scattered all over the RF spectrum.
As soon as there is any legitimate need for these frequencies, I am sure that the history repeats.
Microwave ovens operate in the 2.4 GHz band, and are poorly shielded high power transmitters. Wireless lans and other ISM band applications are low power transmitters, but many microwave ovens (especially older ones) radiate (read: interfere) more power than these low power transmitters are allowed to do, because of their poor shielding. I would be very sceptical to use equipment running in the 2.4 GHz for "mission critical" work. 5.7 GHz ISM band equipment would be a better choice.
Introduction Linux 2.2.16 is the latest update to the Linux kernel tree. The out of the box tree supports the Alpha, PPC, S/390, Sparc and X86 platforms. MIPS ismostly merged but you should obtain the platform specific tree. ARM and M680x0 users should get their platform specific tree.
Compilers This code is intended to build with gcc 2.7.2 and egcs 1.1.2. Patches for building with gcc 2.95 are merged but less tested than other compilers. Caution is recommended when using gcc 2.95 and feedback is sought.
Binary Compatibility Linux 2.2.16 changes a few internal system structures. You may need to rebuild a few third party modules such as pcmcia-cs when upgrading from older kernels to this one.
Security Notes
Linux 2.2.16 is primarily a security release. It includes fixes for both local and network related bugs. Upgrading is strongly recommended.
Security Updates
Capabilities Fixes for serious setuid handling flaws when using restricted capability sets ELF loader The ELF loader could be tricked by erroneous headers Procfs Several/proc drivers failed to do correct sanity checking Readv/writev Potential overflow bug fixed Signal Stacks Exec failed to clear an existing alternate sigstack System 5 Shared Memory If a user managed to attach a segment 65536 times bad things happened. TCP multiconnect hang The TCP code had a bug that could cause the machine to hang. This was user exploitable.
Architecture Updates
Alpha Fix SRM handling Export symbols needed for modular tv card support Fix SMP rescheduling with lock held Handle early Monet boards
i386 Handle IBM thinkpad APM bios again Attempt to work around broken BIOS MP1.4 tables Interrupt controller hanging changed to handle possible buggy chipsets In a few cases IRQ probing was fooled by longstanding pending IRQs Detect and report Intel 'Cascades' series processors Support processors over 4.3GHz in speed
MIPS
PowerPC
S/390 Resynchronized with the IBM code base. Multiple fixes. IBM S/390 partition formats.
Sparc Sparc64 OBP fixup fixes Envctrl driver updates Fix mishandling of some unaligned exceptions Fix tlb flushing bug Sbus audio fixes for poll() Report correct errors on sunmouse errors
Core Updates
Elevator algorithm changes The disk scheduling algorithm is now fair over short as well as long terms Kmod The module loader spots loops and acts sensibly if they occur VM fixes Improve the virtual memory subsystem behaviour
Driver Updates
Adaptec 152x Recognize the AIC6370Q cards ATI frame buffer Fix PCI address handling errors CDROM Generic CD-ROM layer enhancements akin to 2.4test CMPCI audio (CMPCI 83x8) The SP/DIF output is now supported and a DMA bug fixed Computone Serial Updated to rev 1.2.9 Console A memory scribble in the console driver has been cured. CPiA Camera Driver updates and fixes Cyclades Serial Report physical addresses, PLX9050 bug workaround, improved performance for TX on Cyclom-Y Girbil dongle A timing problem with some devices has been fixed I2O Block Support added for dynamic volume creation/deletion I2O core Fix several bugs in the core IDE-CD Remove the ghost DVD hack. DVD-RAM is now writable directly. IDE-CD Unified audio ioctls, packet interfaces using MMC2. Fix possible OOPS IDE-CD Add DVD ioctls needed for DVD movie players IDE Disk Handle drives jumpered for 4092 cylinders IDE Disk Avoid automatic DMA enables on the 450NX IDE Probing Fix a bug that sometimes caused CD-ROM or LS-120 probe errors IDE Recognize Simple Tech ATA Flash disks. INI9100U Handle shared IRQs Intel ICH audio A minimal driver for the i810 audio is now included. ISDN Fix multilink PPP problems Keyboard Handle PS/2 style reconnect code sequences. Lp Added more checks to careful mode Maestro audio Poll bugs have been fixed and a potential crash on unload. MDA console Fix cursor bugs Parallel IDE This now tries to autoload a protocol module Parport Add TIMEDIA 1889 support PSS Audio Joystick support sorted out, cleaned up code and more RAM disk size limit This is now configurable Random driver Remove key repeat codes from random entry pool - they are too predictable SBC-60XX A driver for the watchdog on this board has been added SCSI CD-ROM Removed the GHOST hack. SCSI DVD-RAM are now writable directly. SCSI Disk Driver Correct handling of disks with 4K block sizes SCSI Generic Updated to the current revision Seagate SCSI Recognize the IBM F1 V1.20 card ServeRAID Updated to the 4.0 driver SyncLink Updated to handle Synchronous PPP and Cisco HDLC Trident 4DWave driver New sound driver added. Also supports the SiS 7018 and ALI5451 TTY Layer Return -EFAULT rather than ignoring invalid I/O requests. VGA console Disable the IRQ on the vga frame buffer VIA 82cxxx The driver now supports native mode audio. Yamaha PCI Audio A legacy mode driver has been added. An ALSA native mode driver is in progress.
File System Updates
Ext2fs Fix a long standing but never observed bitmap handling bug FAT Clean up multibyte encoding handlers ISOfs Handle sessions better NCPfs Mixed updates NFS Fix potential machine hang in nfs_free_dentries Partitions Disks with old style partitions on large block sizes are now automatically recognized and handled. SMBfs Assorted updates, removal of debugging messages. POSIX unlink semantics UFS Fixed buffer leak on full disks
Miscellaneous Updates
Configuration Both Menuconfig and Xconfig have been improved. Gcc 3.x Change compiler tests ready for when gcc 3.0 eventually appears
Network Updates
3c515 Fix a bug where the board hung after 2^32 packets 3c59x Extensive updates and bug fixes to this driver. NWAY on the 3c590C 82596 Performance enhancements and more Acenic Updated to 0.44. Fixes for a crash sometimes seen with dhcp clients Appletalk Several cases where appletalk would oops on device downs have been fixed C101 Added a synchronous driver for the Moxa C101 DGRS Support shared IRQ mode. Handle gcc 2.95 builds DHCP DHCP is now supported on diskless boot DMFE Remove surplus error messages EEpro Add support for the older ee10 boards (82595FX etherexpress 10) EEpro100 Workaround FCB interrupt lockups, clean up 82559ER support. Honour PortReset timing. Hamachi Driver for the Packet Engines GNIC-II added IBM TR The windowed shared ram is now supported in full IP Masquerade A memory scribble in the masquerade code has been fixed IPX Fix a memory leak in the IPX layer IRDA Fix for automatic bandwidth setting Olympic TR The IBM PCI adapters now work on LinuxPPC Riscom N2 Support for the RISCom/N2 added. (not the integrated CSU/DSU) SBE WanXL Support for this has been added. SMC9194 Fix board memory allocation bug. SyncPPP Added ioctls for changing flags TCP Fix a crash on certain unusual TCP retransmit patterns Unix sockets Provide credentials on socketpair()
<conspiracy mode> This bug was introduced in PGP 5.0 and fixed in PGP 6.5. Why wasn't this reported on bugtrack, a long time ago ? Although the code is substantially rewritten, I am would be very suprised if the author of this code in 6.5 didn't see this bug (after all he fixed it...) </conspiracy mode>
Bob Auger of Electric Switch, a DVD production company, says: "This is the first time DVD is being seen as it is meant to be seen."
This is just wrong ! He is showing DVD in SDI quality which basically is digital PAL. PAL has a resolution of 625x360 pixels. DVD is encoded with 720x480 pixels. He is using 225000 pixels out of 345600....
The DVD format is so advanced, it does not exit any player which uses all features (yet). This is a short list:
Over 2 hours of high-quality digital video (over 8 on a double-sided, dual-layer disc).
Support for widescreen movies on standard or widescreen TVs (4:3 and 16:9 aspect ratios).
Up to 8 tracks of digital audio (for multiple languages, DVS, etc.), each with as many as 8 channels.
Up to 32 subtitle/karaoke tracks.
Automatic "seamless" branching of video (for multiple story lines or ratings on one disc).
Up to 9 camera angles (different viewpoints can be selected during playback).
Menus and simple interactive features (for games, quizzes, etc.).
Multilingual identifying text for title name, album name, song name, cast, crew, etc.
Instant rewind and fast forward (no "be kind, rewind" stickers and threats on rental discs)
Instant search to title, chapter, music track, and timecode.
Durable (no wear from playing, only from physical damage).
Not susceptible to magnetic fields. Resistant to heat.
Compact size (easy to handle, store, and ship; players can be portable; replication is cheaper).
Noncomedogenic.
I think that the ability to select up to 9 different camera angles during playback is one of the cooles features !
This software is not a toy, it is one of the best pcb design packages available on any platform, and it is reasonably priced. Life is good to us electronics engineers !
Microwindows/NanoGUI is an Open Source project aimed at bringing the features of modern graphical windowing environments to smaller devices and platforms. NanoGUI allows applications to be built and tested on the Linux desktop, as well as cross-compiled for the target device.
Both share a common graphics engine. Nano-GUI is based on an X-like protocol called Nano-X. Microwindows sports an interface similar to the ECMA APIW spec with some advancements.
As always, Matrox has excellent linux support. G400 runs fast and steady under XFree86 SVGA server. You can overclock the board with gMGAclock, a GNOME-based overclocking utility for Matrox G400 cards.
The idea of the plex86 project is to create a viritual machine which analyzes the code, and determines which parts of the code that can run safely on the CPU without touching the protection mechanism. This requiers heavy cooperation between the paging/protection mechanisms of the two opertating systems, and a minor change in one of them, requiers changes in freevmware. This is why vmware need to know which OS to run as guest OS.
Then came Transmeta and Crusoe. This CPU has "hardware" just in time compilation of X86 code, which means that the cpu compiles the x86 code to Crusoe native code before executing, and saves the compiled code in a cache. The cpu discards old compiled code in the cache, which stops the cache from growing beyond limits.
These techniques could be combined, by doing jit compiling of x86 code to x86 code (ie. coping the code to a cache and adding breakpoints to the compiler at every instruction which isn't compiled before or discarded from the cache). This is fast, since no "real" compilation is necessary. On the Crusoe the compilation is much harder since you compile to another instruction set. If this is possible, running a guest OS on a linux machine should be very fast and smooth, and all this could be done in user mode, without kernel support. As more of the guest os code is "compiled", we could discard (or page out) the "interpreted" code (the original os code) thus saving memory for the cache. The memory usage of vmware has disappointed more than one user.
Sharing IRQ's between PCI and ISA does not work, which implies that your SCSI controller might be forced to share an interrupt or dma channel with one of the the ethernet controllers if you have a ISA card in your system. If the card had been a PCI card it could share the IRQ with the SCSI controller for the backup streamer, which would have minor impact on the system performance.
In a PC, every important card (scsi, ethernet, etc.) should have its own unshared interrupt. Adding ISA cards for low bandwidth purposes, eats interrupts from the high bandwidth tasks.
Parity shouldn't be an issue, since a bus isn't supposed to fail. The mechanism exists to isolate errors, since debugging hardware incompatibility is very difficult.
The industry chooses lower supply voltages on electronics simply because it delivers better price/performance ratio, but this does not apply it you have to add a switch mode power supply to your ISA card to create the +3.3 V from +5V.
So in short: Why would anyone in their right mind make a new design based on the ISA bus ?
Why would anyone in their right mind make a new design based on the ISA bus ? This is a brain dead bus, without interrupt sharing and severe limitations on DMA and IO. There does not exist a complete specification on the bus, only a collection of random writings and books. There was an attempt by ieee (?) to make a spec, but they failed because of the ugliness of the existing implementations. It has an even more kludgey and unreliable plug and pray specification. The bus has no error correction, parity or ECC. I refuse to put any ISA card in any new computer (cheap home pc or desktop for the secretary), and this card is meant for servers !!
This card is a software raid solution, the flash prom contains software raid drivers similar to the linux md drivers. This modification makes sense if you need raid under dos/windows which is supported by Promises software bios and drivers. Linux has its own drivers doing the same thing, called md.
The resistor (yep, it's a resistor, not a transistor) is only in place to distinguish between the two boards, and make sure that users who need software raid in bios have to pay for it.
There is no reason why a modified card should give higher performance under linux what so ever.
Slashdot Mirror
Hmmm, 9A in 10 sec = 25 mAh, that is more than the 20 mAh battery capacity. There is something wrong here !
Buy the PC without a harddrive, and buy the harddrive separate. I did that when I bought my laptop which of course is Micros~1 free. I didn't even have to return any CD with that legacy OS !!
This is silly ....
I did a quick survey on the "Norman virus scanner, it knows about 18.284 viruses, that is more than 25% of all programs written according to the "Cato Institute". That simply has to be wrong, their estimate of 70.000 programs is way to low.
> 70 dBm is 70 dB (=10000000 times) above 1 mW = 10 W.
> 70 dBW is 70dB above 1 W = 10kW.
Gee, where was I when I wrote that ?
Yes, 70dBm is 10kW, but it only makes the case more clear. By using VMSK/2 modulation you use a lot of power to conserve bandwidth. This can be advantageous in some situations, but clearly *not* when you need high baud rates
70 dBm is 70 dB (=10000000 times) above 1 mW = 10 W.
...
70 dBW is 70dB above 1 W = 10kW.
I think the math is OK
Nice theory, but this is impossible to implement:
....
Do the math:
Shannon's channel capacity theorem:
C= B * log2 ( 1 + SNR )
C = capacity in bps
B = bandwidth
SNR = signal to noise ratio
Solving for SNR in dB ( = 10*log10(SNR) ) gives:
SNR_db = 10 * log10 (( 2 ^ (C/B)) - 1)
With C = 12.8 Mbit/s and B = 200 kHz you get SNR = 192 dB !
To have a SNR = 192 dB, the signal has to be 1.8 * 10^19 times stronger than the noise ! A receiver with 200kHz bandwidth will typically have a noise floor at -120 dBm, so you need more than 70 dBm received signal strength. 70 dBm is 10 W !! And that is the signal strength at the receiving antenna, so the transmitter would have to be in the gigawatt range, to reach short distances.
This means that VMSK/2 can be used, but you can't reach 12.8 Mbit/s without a nuclear powered transmitter. You can get a decent bit rate with VMSK/2 on battery powered equipment, but you have to design for a few kbit/s, not 12.8 Mbit/s. Nice theory, but
Read my lips
Does this mean that RH7.0 will be a final snapshot from rawhide, and not a upgraded 7.0beta ? In a few weeks when the bug reports starts ticking in, will rawhide be more stable than 7.0beta, or is rawhide always the bleeding edge unstable redhat where we can crashtest all the new stuff ?
Cheops is a network "swiss army knife". It's "network neighborhood" done right (or gone out of control, depending on your perspective). It seems that the development has slowed down a bit though.
Have a look at:
http://www.marko.net/cheops/and
http://www.marko.net/cheops/features.htmlThis technology does not affect your privacy !
It will only affect you if:
Itrace will only give useful information about users who are abusing the flaws in the tcp/ip protocol and at the same time are sending lots of data (like syn flooding), and will not have any consequences for the rest of us.
I like simple protocols/standards with no possibilities for extensions. Look at history:
Kerberos: It has a field which is reserved. The standard was abused by MS, and the extension is now a "trade secret" making "MS-Kerberos" a property protocol.
SMTP mail: Can be extended with X- headers. Used is many incompatible manners by diverse email software.
HTML: A standard which is designed to be easy to extend. Need I say more ?
Keep the protocols simple with no possibilities for extensions, and make sure that implementations don't fail gracefully when abused. If so, all implementations will work together, and MS and other will fail when they apply their "embrace and extend" tactics.It could not have been more wrong, and today the radio amateurs are crammed into small segments scattered all over the RF spectrum.
As soon as there is any legitimate need for these frequencies, I am sure that the history repeats.
Microwave ovens operate in the 2.4 GHz band, and are poorly shielded high power transmitters. Wireless lans and other ISM band applications are low power transmitters, but many microwave ovens (especially older ones) radiate (read: interfere) more power than these low power transmitters are allowed to do, because of their poor shielding. I would be very sceptical to use equipment running in the 2.4 GHz for "mission critical" work. 5.7 GHz ISM band equipment would be a better choice.
Linux 2.2.16 Release Notes
/proc drivers failed to do correct sanity checking
Platforms:Alpha, PowerPC, S/390, Sparc, X86
Introduction
Linux 2.2.16 is the latest update to the Linux kernel tree. The out of the box tree supports the Alpha, PPC, S/390, Sparc and X86 platforms. MIPS ismostly merged but you should obtain the platform specific tree. ARM and M680x0 users should get their platform specific tree.
Compilers
This code is intended to build with gcc 2.7.2 and egcs 1.1.2. Patches for building with gcc 2.95 are merged but less tested than other compilers. Caution is recommended when using gcc 2.95 and feedback is sought.
Binary Compatibility
Linux 2.2.16 changes a few internal system structures. You may need to rebuild a few third party modules such as pcmcia-cs when upgrading from older kernels to this one.
Security Notes
Linux 2.2.16 is primarily a security release. It includes fixes for both local and network related bugs. Upgrading is strongly recommended.
Security Updates
Capabilities
Fixes for serious setuid handling flaws when using restricted capability sets
ELF loader
The ELF loader could be tricked by erroneous headers
Procfs
Several
Readv/writev
Potential overflow bug fixed
Signal Stacks
Exec failed to clear an existing alternate sigstack
System 5 Shared Memory
If a user managed to attach a segment 65536 times bad things happened.
TCP multiconnect hang
The TCP code had a bug that could cause the machine to hang. This was user exploitable.
Architecture Updates
Alpha
Fix SRM handling
Export symbols needed for modular tv card support
Fix SMP rescheduling with lock held
Handle early Monet boards
i386
Handle IBM thinkpad APM bios again
Attempt to work around broken BIOS MP1.4 tables
Interrupt controller hanging changed to handle possible buggy chipsets
In a few cases IRQ probing was fooled by longstanding pending IRQs
Detect and report Intel 'Cascades' series processors
Support processors over 4.3GHz in speed
MIPS
PowerPC
S/390
Resynchronized with the IBM code base. Multiple fixes.
IBM S/390 partition formats.
Sparc
Sparc64 OBP fixup fixes
Envctrl driver updates
Fix mishandling of some unaligned exceptions
Fix tlb flushing bug
Sbus audio fixes for poll()
Report correct errors on sunmouse errors
Core Updates
Elevator algorithm changes
The disk scheduling algorithm is now fair over short as well as long terms
Kmod
The module loader spots loops and acts sensibly if they occur
VM fixes
Improve the virtual memory subsystem behaviour
Driver Updates
Adaptec 152x
Recognize the AIC6370Q cards
ATI frame buffer
Fix PCI address handling errors
CDROM
Generic CD-ROM layer enhancements akin to 2.4test
CMPCI audio (CMPCI 83x8)
The SP/DIF output is now supported and a DMA bug fixed
Computone Serial
Updated to rev 1.2.9
Console
A memory scribble in the console driver has been cured.
CPiA Camera
Driver updates and fixes
Cyclades Serial
Report physical addresses, PLX9050 bug workaround, improved performance for TX on Cyclom-Y
Girbil dongle
A timing problem with some devices has been fixed
I2O Block
Support added for dynamic volume creation/deletion
I2O core
Fix several bugs in the core
IDE-CD
Remove the ghost DVD hack. DVD-RAM is now writable directly.
IDE-CD
Unified audio ioctls, packet interfaces using MMC2. Fix possible OOPS
IDE-CD
Add DVD ioctls needed for DVD movie players
IDE Disk
Handle drives jumpered for 4092 cylinders
IDE Disk
Avoid automatic DMA enables on the 450NX
IDE Probing
Fix a bug that sometimes caused CD-ROM or LS-120 probe errors
IDE
Recognize Simple Tech ATA Flash disks.
INI9100U
Handle shared IRQs
Intel ICH audio
A minimal driver for the i810 audio is now included.
ISDN
Fix multilink PPP problems
Keyboard
Handle PS/2 style reconnect code sequences.
Lp
Added more checks to careful mode
Maestro audio
Poll bugs have been fixed and a potential crash on unload.
MDA console
Fix cursor bugs
Parallel IDE
This now tries to autoload a protocol module
Parport
Add TIMEDIA 1889 support
PSS Audio
Joystick support sorted out, cleaned up code and more
RAM disk size limit
This is now configurable
Random driver
Remove key repeat codes from random entry pool - they are too predictable
SBC-60XX
A driver for the watchdog on this board has been added
SCSI CD-ROM
Removed the GHOST hack. SCSI DVD-RAM are now writable directly.
SCSI Disk Driver
Correct handling of disks with 4K block sizes
SCSI Generic
Updated to the current revision
Seagate SCSI
Recognize the IBM F1 V1.20 card
ServeRAID
Updated to the 4.0 driver
SyncLink
Updated to handle Synchronous PPP and Cisco HDLC
Trident 4DWave driver
New sound driver added. Also supports the SiS 7018 and ALI5451
TTY Layer
Return -EFAULT rather than ignoring invalid I/O requests.
VGA console
Disable the IRQ on the vga frame buffer
VIA 82cxxx
The driver now supports native mode audio.
Yamaha PCI Audio
A legacy mode driver has been added. An ALSA native mode driver is in progress.
File System Updates
Ext2fs
Fix a long standing but never observed bitmap handling bug
FAT
Clean up multibyte encoding handlers
ISOfs
Handle sessions better
NCPfs
Mixed updates
NFS
Fix potential machine hang in nfs_free_dentries
Partitions
Disks with old style partitions on large block sizes are now automatically recognized and handled.
SMBfs
Assorted updates, removal of debugging messages. POSIX unlink semantics
UFS
Fixed buffer leak on full disks
Miscellaneous Updates
Configuration
Both Menuconfig and Xconfig have been improved.
Gcc 3.x
Change compiler tests ready for when gcc 3.0 eventually appears
Network Updates
3c515
Fix a bug where the board hung after 2^32 packets
3c59x
Extensive updates and bug fixes to this driver. NWAY on the 3c590C
82596
Performance enhancements and more
Acenic
Updated to 0.44. Fixes for a crash sometimes seen with dhcp clients
Appletalk
Several cases where appletalk would oops on device downs have been fixed
C101
Added a synchronous driver for the Moxa C101
DGRS
Support shared IRQ mode. Handle gcc 2.95 builds
DHCP
DHCP is now supported on diskless boot
DMFE
Remove surplus error messages
EEpro
Add support for the older ee10 boards (82595FX etherexpress 10)
EEpro100
Workaround FCB interrupt lockups, clean up 82559ER support. Honour PortReset timing.
Hamachi
Driver for the Packet Engines GNIC-II added
IBM TR
The windowed shared ram is now supported in full
IP Masquerade
A memory scribble in the masquerade code has been fixed
IPX
Fix a memory leak in the IPX layer
IRDA
Fix for automatic bandwidth setting
Olympic TR
The IBM PCI adapters now work on LinuxPPC
Riscom N2
Support for the RISCom/N2 added. (not the integrated CSU/DSU)
SBE WanXL
Support for this has been added.
SMC9194
Fix board memory allocation bug.
SyncPPP
Added ioctls for changing flags
TCP
Fix a crash on certain unusual TCP retransmit patterns
Unix sockets
Provide credentials on socketpair()
>
/* burn it.*/
...)
> Versions 2.* and 6.5 of PGP do NOT share this problem.
>
This is how this was fixed in pgp 6.5i:
if((fdesc=open( devrandom, O_RDONLY|O_NONBLOCK)) > 0) {
while((numread = read( fdesc, buffer, BUFFSIZE)) > 0) {
for( p = buffer ; numread > 0 ; numread--, p++ ) {
PGPGlobalRandomPoolAddKeystroke(*p);
*p=0;
}
RandBits = PGPGlobalRandomPoolGetEntropy();
StillNeeded = TotalNeeded - RandBits;
}
}
<conspiracy mode>
This bug was introduced in PGP 5.0 and fixed in PGP 6.5. Why wasn't
this reported on bugtrack, a long time ago ? Although the code is
substantially rewritten, I am would be very suprised if the author
of this code in 6.5 didn't see this bug (after all he fixed it
</conspiracy mode>
"This is the first time DVD is being seen as it is meant to be seen."
This is just wrong ! He is showing DVD in SDI quality which basically is digital PAL. PAL has a resolution of 625x360 pixels. DVD is encoded with 720x480 pixels. He is using 225000 pixels out of 345600....
The DVD format is so advanced, it does not exit any player which uses all features (yet). This is a short list:
I think that the ability to select up to 9 different camera angles during playback is one of the cooles features !
Two years is extremely slow in the computer age ....
http://www.linux.org.uk/VERSION/ relnotes.2211.html
http://www.linux.org.uk/VERSION/ relnotes.2212.html
http://www.linux.org.uk/VERSION/ relnotes.2213.html
http://www.linux.org.uk/VERSION/ relnotes.2214.html
Do you see the pattern here ? The changelog for 2.2.15 will most likely be available at:
http://www.linux.org.uk/VERSION/ relnotes.2215.html
Be patient, have fun :-)
They even have a freeware version which is ok for weekend projects.
http://www.cadsoftusa.com/ or http://www.cadsoft.de/
This software is not a toy, it is one of the best pcb design packages available on any platform, and it is reasonably priced. Life is good to us electronics engineers !
Both share a common graphics engine. Nano-GUI is based on an X-like protocol called Nano-X. Microwindows sports an interface similar to the ECMA APIW spec with some advancements.
As always, Matrox has excellent linux support. G400 runs fast and steady under XFree86 SVGA server. You can overclock the board with gMGAclock, a GNOME-based overclocking utility for Matrox G400 cards.
Then came Transmeta and Crusoe. This CPU has "hardware" just in time compilation of X86 code, which means that the cpu compiles the x86 code to Crusoe native code before executing, and saves the compiled code in a cache. The cpu discards old compiled code in the cache, which stops the cache from growing beyond limits.
These techniques could be combined, by doing jit compiling of x86 code to x86 code (ie. coping the code to a cache and adding breakpoints to the compiler at every instruction which isn't compiled before or discarded from the cache). This is fast, since no "real" compilation is necessary. On the Crusoe the compilation is much harder since you compile to another instruction set. If this is possible, running a guest OS on a linux machine should be very fast and smooth, and all this could be done in user mode, without kernel support. As more of the guest os code is "compiled", we could discard (or page out) the "interpreted" code (the original os code) thus saving memory for the cache. The memory usage of vmware has disappointed more than one user.
In a PC, every important card (scsi, ethernet, etc.) should have its own unshared interrupt. Adding ISA cards for low bandwidth purposes, eats interrupts from the high bandwidth tasks.
Parity shouldn't be an issue, since a bus isn't supposed to fail. The mechanism exists to isolate errors, since debugging hardware incompatibility is very difficult.
The industry chooses lower supply voltages on electronics simply because it delivers better price/performance ratio, but this does not apply it you have to add a switch mode power supply to your ISA card to create the +3.3 V from +5V.
So in short: Why would anyone in their right mind make a new design based on the ISA bus ?
Green engineers ?
Why would anyone in their right mind make a new design based on the ISA bus ? This is a brain dead bus, without interrupt sharing and severe limitations on DMA and IO. There does not exist a complete specification on the bus, only a collection of random writings and books. There was an attempt by ieee (?) to make a spec, but they failed because of the ugliness of the existing implementations. It has an even more kludgey and unreliable plug and pray specification. The bus has no error correction, parity or ECC. I refuse to put any ISA card in any new computer (cheap home pc or desktop for the secretary), and this card is meant for servers !!
Oh, did I mention that the bus is +5V only ?
Oh well, back to work.
</rant>
This card is a software raid solution, the flash prom contains software raid drivers similar to the linux md drivers. This modification makes sense if you need raid under dos/windows which is supported by Promises software bios and drivers. Linux has its own drivers doing the same thing, called md.
The resistor (yep, it's a resistor, not a transistor) is only in place to distinguish between the two boards, and make sure that users who need software raid in bios have to pay for it.
There is no reason why a modified card should give higher performance under linux what so ever.