wavemon and tinycore kernel

[gerrit]

A user reported two problems with wavemon, one of them I could fix

• no informative help (-h/-v) if no interfaces found<br>
      This is fixed in eden-feed.erg.abdn.ac.uk/wavemon/wavemon-current.tar.bz2
      which is based on the current stable release 0.7.3
  
• wirelss interfaces are loaded (show up in iwconfig), but are not present in /proc/net/wireless

The second issue was the original problem. wavemon is (like iwlist and iwconfig) based on the wireless
extensions. Testing with the Core Plus 4.2 distribution, the iwl3945 network interface was (including the
firmware) loaded, but was not listed in /proc/net/wireless. It could therefore not be found by wavemon,
which complained that no wireless interfaces were found.
I had a quick look over the 3.0.3 kernel .config, it has the WEXT/WIRELESS_EXT enabled, but it could be
that there is just one more tweak necessary to bring the wireless interfaces back into /proc/net/wireless

Could someone please have a look at this issue?

Here is the output from a 3.2.0-rc7 kernel with the same laptop  and iwl3945 driver:

Code: (bash) [Select]

Inter-| sta-|   Quality        |   Discarded packets               | Missed | WE
 face | tus | link level noise |  nwid  crypt   frag  retry   misc | beacon | 22
 wlan0: 0000    0     0     0        0      0      0      0      0        0

[bmarkus]

According to your post it has nothing to do with Wavemon. Your subject is misleading.

[curaga]

CONFIG_WEXT_PROC=y which should handle that. Perhaps a bug instead of a config option miss.

[gerrit]

Sorry for the late reply. The subject was misleading since I was not clear what exactly the cause had been. The described setting for /proc
is in the tiny core configuration. Found the cause of the problem by recompiling a kernel using the tinycore 3.0.3 configuration.

The cause for wavemon not being able to find the interfaces is that /proc/net/wireless may exist, but may not list all wireless interfaces.
I have fixed this problem in the current development release of wavemon on eden-feed.erg.abdn.ac.uk/wavemon/wavemon-current.tar.bz2
Tested and verified that this fix works with the tinycore configuration as well as with other configurations.

That fix is needed to make wavemon work on tinycore linux; the package for wavemon should be updated.
(If it is of any help, there will be a new stable release of wavemon (0.7.4) containing this fix Monday next week.)

Gerrit

[bmarkus]

Hi Gerrit,

thanks for the info. I'm preferring to wait stable release and then update package.

Regards... Béla

[gerrit]

Since last week I have done some more testing and put out a new stable release, 0.7.4 now on

eden-feed.erg.abdn.ac.uk/wavemon/stable-releases/wavemon-0.7.4.tar.bz2

This should fix the issues reported on Tiny Core 0.4.0 and also observed with the 3.0.3 kernel configuration.
There are also a few minor improvements (search ordering in scan window).

Best regards
Gerrit

[bmarkus]

Gerrit

Just tested 0.7.2 from the repo with 4.4.rc2 (3.0.21) and it works correctly. The only issue I found that F1/F2/F3 keys doesn't work at least on the stock TC, F7/F8/F9/F10 are ok. Seems to be an aterm issue, in Lxterm fkeys are OK except F10 which conflicts with lxterm's own quit. Adding (configurable?) short keys would solve the problem with various terminals.

0.7.4 behaves on the same way. I'm submitting it to the repo.

Béla

[gerrit]

Bela,

thank you for reporting the key-binding issue. I will see what I can do. ( An earlier overlap with gnome-terminal's F1
help key had lead to adding the short-cut 'i' for the info screen.)
The fix in 0.7.4 is for the following problem: if /proc/net/wireless is empty, wavemon was not able to find the wireless
interfaces and would not even start. If it now encounters an empty /proc/net/wireless, it will no longer affect operation,
since /proc/net/dev is used instead. If really no wireless interfaces exist at all, it will report this and exit.

Gerrit

[gerrit]

Update on the aterm/lxterm problem:

I have added two fixes to address the problems described above, both are in the development tarball (URL
as above), on top of 0.7.4.

The F1..F4 keys on aterm produce different sequences than advertised by the rxvt terminfo entry, it produces
the vt100 PF1..4 \EOP..\EOS key sequences. Due to the mismatch, the key sequences are sent uninterpreted,
so that the F1..F4 keys are not recognized (htop has the same problem, mc works around this).

The solution to tackle this problem is to catch those escape sequences and to re-interpret them as F1..F4 keys.
It was tested to work with aterm,  it may also work with similar terminals.

For the other problem of terminals that remap F1..10 there is a second fix: each screen now additionally
has a shortcut corresponding to the first letter of the screen name
(F1 -> 'i', F2 -> 'l', F3 -> 's', F7 -> 'p', F8 -> 'h', F9 -> 'a', F10 -> 'q').

It is probably too early to do another stable release, since this only has two new patches on top of 0.7.4, would it
be possible to test the above update -- I believe it addresses the described problems.

[bmarkus]

Just tested the new version and it works fine, so Iwill update TC repo keeping version number 0.7.4 and indicating changes in .info