Martin, Unless you or somebody else really need it, I am not planning to add more backward compatibility to the "7mdv" tarball than was originally available in the "4mdv" sources. I think "4mdv" didn't support the 2.6.26 kernel, while "2mdv" did. I think Zsolt has made some progress with the "8mdv"/"9mdv" sources available at his site [http://linux.zsolttech.com/linmodem/agrsm/Testing/]. IIRC his packages support a wider range of kernels but they need some testing too. Search the list archive for his message on the subject: <<<< Date: Wed, 3 Jun 2009 22:29:34 +0200 From: Zsolt László BÉRCZES <zsolt@xxxxxx> To: Discuss@xxxxxxxxxxxxx Subject: new packages (was: agrsm on 2.6.30) >>>> In the meantime, the change that you can try for the "7mdv" driver under the 2.6.26 kernel is this one: in serial26.c, line 990, change struct tty_struct *tty = up->port.info->port.tty; to struct tty_struct *tty = up->port.info->tty; It may be enough to get the "7mdv" working under 2.6.26, or it may be not. I didn't have enough time to investigate it any further. -- WBR, Nikolay Zhuravlev Thus spake Martin Lupcke (mossianin_lbcmi@xxxxxxxxxx): > Forgive my delay. Too much going on. > > Nikolay mentioned in this note that he was going to make a change to one > of the "mdv" (I believe 7mdv) series files and make it available. Can > someone tell me if this is done and what the file is? > > Thanks. > > > Nikolay Zhuravlev wrote: >> All, >> First, a correction. The patches I supplied didn't break the >> backward compatibility in this particular case. The patches >> were against the "4mdv" tarball available at the Zsolt's site. >> The "4mdv" has the serial26.c file with the following code >> around the line 990: >> >> <<<< >> receive_chars(struct uart_8250_port *up, int *status, struct pt_regs *regs) >> { >> struct tty_struct *tty = up->port.info->port.tty; >> >> >> The older kernels (< 2.6.26) will not compile that, because they have a >> different definition of the struct uart_info. Those interested in the older >> kernels may have better luck with the "2mdv" sources (agrsm-2.1.80-2mdv2009.0.src.rpm). >> This is how the same code looks in the "2mdv" serial26.c: >> >> <<<< >> receive_chars(struct uart_8250_port *up, int *status, struct pt_regs *regs) >> { >> struct tty_struct *tty = up->port.info->tty; >> >> >> Second, a clarification. The patches I posted to the list do not add support for >> more chipsets. They 1) do a little clean up of the source build system, 2) get >> rid of some warning messages that clutter the output, and 3) make it so that the >> sources build successfully for the more recent kernels. >> >> Note also that "build successfully" does not necessarily mean "function >> properly". I've built them against 2.6.{27,28,29} and verified that wvdialconf >> reports success. I didn't do any more testing beyond that. >> >> So, if the "2mdv"/"4mdv" sources didn't support your chipsets, the "7mdv" is not >> likely to support them either. >> >> Testing was needed to verify that the "7mdv" didn't break anything that was >> working before in the "2mdv"/"4mdv" sources. Also, Marvin injected the 630 (or >> was it 620?) chipset along with my patch in the hopes that it might just work >> [as it sometimes do] but it didn't. >> >> Martin, >> >> The patch is not coming yet, I will get back to this issue in a few days. >> >>
