p54: Fix compilation problem on PPC

The commit entitled "p54: Fix sparse warnings" introduced a compile error on PPC architecture. Thanks to Johannes Berg <[email protected]> for reporting this problem.

---

John,

This is 2.6.28 material. I hope this comes through OK. I don't have access to my usual way of submitting patches.

Larry
---

Johannes,

Do you have access to a p54 device, or were you just compile testing? In my mind, there is still a big question regarding the endianness of one of the data arrays.

Larry
---

Index: wireless-testing/drivers/net/wireless/p54/p54common.c
===================================================================
@@ -482,7 +482,6 @@ static int p54_parse_eeprom(struct ieee8
 	printk(KERN_ERR "p54: eeprom parse failed!\n");
 	return err;
 }
-EXPORT_SYMBOL_GPL(p54_parse_eeprom);
 
 static int p54_rssi_to_dbm(struct ieee80211_hw *dev, int rssi)
 {

No, I don't have a device, was just compiling, sorry. Is there any on the market that isn't too expensive that I could get?

johannes

This is the one I have:

http://cgi.ebay.de/Dell-wireless-1450-USB-adapter-NEW-in-box_W0QQitemZ170268289125QQcmdZViewItem?hash=item170268289125&_trkparms=72%3A1282|39%3A1|66%3A2|65%3A12|240%3A1318&_trksid=p3286.c0.m14

The current bid is 7,40 Eur.

Larry

I'm sorry for being late. I have tested an Xterasys XG-600 (miniPCI) on i386 and PowerPC. The same device is working on i386, but not on PowerPC. There are kernel messages running on the console on PowerPC, but I could not capture them yet. The kernel is the current wireless-testing.

I also tested D-Link DWL-G650 (CardBus), but it appears to be unreliable. I could scan only once with it on i386. Most time scanning won't find any APs.

I suspect we should not be mixing little-endian and host-endian structures.

Regards,
Pavel Roskin

Agreed. In data structure "bootrec", there is a data array that gets used once as be32, once as le16, and several times as though it were in native CPU order. That has always look funny to me, but as I can only test on x86_64 architecture, I cannot distinguish le32 from native CPU.

I will prepare a patch to explicitly use those data as le32 for you to test.

Thanks,

Larry