I am unable to see the behavior I saw before: namely ifconfig rxcount goes to 0, txcount goes low, radio off, radio on, rxcount+, txcount+++, dhcp works, give it a second, down again. repeats.
Right now what I'm seeing is a perfectly working system. I have moved different APs within the same ESSID (WEP enabled) and no problems are evident. I'm getting TCP throughput of 7Mbps (on my Comcast 6Mbps link) and I haven't configured any in-LAN testing facilities (yet. The night is young.)
EhudPS I know that newbies get the respect of somewhere between Peter who cried Wolf and zero. I swear it really (repeatedly) didn't work, and now with that one (insignificant...) change it works perfectly. The reboot is not a factor as I'd done three on the old system. (Golden Nugget. Overnight. McCarran. Home.)
Pavel Roskin wrote:
On Mon, 2007-04-09 at 10:08 -0500, Larry Finger wrote:John H. wrote:I just put a standalone tarball for the latest version of bcm43xx-softmac onto my FTP site at ftp://lwfinger.dynalias.org/patches/bcm43xx-softmac-sa.tar.bz2. You should download it, unpack it, cd to bcm43xx-softmac-sa, issue a make command, and finish with a 'sudo make install'. That should build and install new versions of bcm43xx.ko and the various parts of ieee80211 and ieee80211-softmac. The make files have not been thoroughly tested, and I welcome your comments and/or difficulties.I ended up just using the following... http://www.howtoforge.com/kernel_compilation_fedora Can I get a patched version of the bcm43xx source and just compile that?Just a quick feedback on that package. Please don't include Module.symvers, your version is not useful for anyone else. It will be generated is needed. I'm getting a compile error: /home/proski/src/bcm43xx-softmac-sa/bcm43xx/bcm43xx.h:896:3: error: #error "Using neither DMA nor PIO? Confused..."I think the best solution would be to define both CONFIG_BCM43XX_DMA and CONFIG_BCM43XX_PIO. You make want to make itconfigurable, but then don't compile bcm43xx_pio.c and bcm43xx_dma.c unconditionally. The debugfs requirement seems to defeat the whole purpose of the standalone distribution. BCM43xx_NR_LOGGED_XMITSTATUS is defined only if CONFIG_BCM43XX_DEBUG is defined, but bcm43xx_debugfs.c uses it unconditionally. There may be more compile problems, but I gave up at this point. Speaking of the build system, you may want to have a warning in case if anything that is compiled as a module (either bcm43xx or softmac) is already available and linked into the kernel. Make sure the checks don't affect "make clean" and similar maintenance commands.
smime.p7s
Description: S/MIME Cryptographic Signature
_______________________________________________ Bcm43xx-dev mailing list Bcm43xx-dev@lists.berlios.de https://lists.berlios.de/mailman/listinfo/bcm43xx-dev
