Skip to contentSkip to conversation

dbparse: remove usable bandwidth check

PATCH In review Johannes Berg 2 participants 5 messages Dec 4, 2012

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 dbparse: remove usable bandwidth check

From: Johannes Berg <[email protected]>

The kernel uses the bandwidth this way: any frequency rule that is required for any of the 20 MHz subchannels of a given channel must allow the total bandwidth. Therefore, this check is wrong -- a frequency rule may need to be specified with higher bandwidth than it has to allow to form e.g. HT40 out of subchannels.

---
John, this patch needs to be applied before the other one so the new db.txt will be parsed successfully.

 dbparse.py | 3 ---
 1 file changed, 3 deletions(-)
dbparse.py +0 -3
index 893d64e..32cbce9 100755
@@ -162,9 +162,6 @@ class DBParser(object):
                 self._syntax_error("Inverted freq range (%d - %d)" % (start, end))
             if start == end:
                 self._syntax_error("Start and end freqs are equal (%d)" % start)
-            if end - start < bw:
-                self._syntax_error("Invalid bandwidth: %d width channel "
-			"cannot possibly fit between %d - %d" % (bw, start, end))
         except ValueError:
             self._syntax_error("band must have frequency range")
 
Signature · 8 lines
1.8.0



_______________________________________________
wireless-regdb mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/wireless-regdb
Acked-by on v1 1/1
#1 lore raw
Quoted 10 lines · Johannes Berg

From: Johannes Berg <[email protected]>

The kernel uses the bandwidth this way: any frequency rule that is required for any of the 20 MHz subchannels of a given channel must allow the total bandwidth. Therefore, this check is wrong -- a frequency rule may need to be specified with higher bandwidth than it has to allow to form e.g. HT40 out of subchannels.

I'd add into the commit log:

The current issue was spotted when dealing with the world regulatory domain which had overlapping frequency rules -- that is you had to evaluate two ranges to determine if you can use HT40 channel. The primary HT40 channel would fit into one freq range and the subchannel would fit into another frequency band.

An additional change is possible however and that is to have the kernel check not for 40 MHz requirement on subchannels but only 20 MHz but this alone could lead to secondary issues by the current implicit architecture support of overlapping frequencies which are we currently evaluating.

Luis

_______________________________________________
wireless-regdb mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/wireless-regdb

#2 lore raw
In reply to Johannes Berg
Quoted 8 lines · Johannes Berg

From: Johannes Berg <[email protected]>

The kernel uses the bandwidth this way: any frequency rule that is required for any of the 20 MHz subchannels of a given channel must allow the total bandwidth. Therefore, this check is wrong -- a frequency rule may need to be specified with higher bandwidth than it has to allow to form e.g. HT40 out of subchannels.

This patch, and the other patch, are because we currently interpret the rules this way:

http://johannes.sipsolutions.net/files/chans-old.pdf

johannes

_______________________________________________
wireless-regdb mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/wireless-regdb

#3 lore raw
Quoted 10 lines · Johannes Berg
Quoted 8 lines · Johannes Berg

From: Johannes Berg <[email protected]>

The kernel uses the bandwidth this way: any frequency rule that is required for any of the 20 MHz subchannels of a given channel must allow the total bandwidth. Therefore, this check is wrong -- a frequency rule may need to be specified with higher bandwidth than it has to allow to form e.g. HT40 out of subchannels.

Quoted 4 lines

This patch, and the other patch, are because we currently interpret the rules this way:

http://johannes.sipsolutions.net/files/chans-old.pdf

Let me attach that for the archives ...

johannes

_______________________________________________
wireless-regdb mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/wireless-regdb

#4 lore raw
Quoted 11 lines · Johannes Berg
Quoted 10 lines · Johannes Berg
Quoted 8 lines · Johannes Berg

From: Johannes Berg <[email protected]>

The kernel uses the bandwidth this way: any frequency rule that is required for any of the 20 MHz subchannels of a given channel must allow the total bandwidth. Therefore, this check is wrong -- a frequency rule may need to be specified with higher bandwidth than it has to allow to form e.g. HT40 out of subchannels.

Quoted 6 lines
Quoted 4 lines

This patch, and the other patch, are because we currently interpret the rules this way:

http://johannes.sipsolutions.net/files/chans-old.pdf

Let me attach that for the archives ...

And give an example :-)

HT40+ on channel 9 uses the range covered by channels 9 and 13. Channel 13 falls into the second freqband, so that must have @40 because in the kernel we currently check that any freqband that is required for channel 9 or 13 (in this example) allows enough bandwidth for the whole channel.

johannes

_______________________________________________
wireless-regdb mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/wireless-regdb