[Bug 5848] pcmcia novatel merlin u530 not working properly on 2.6.13 and up

[bugme-daemon at bugzilla.kernel.org]

http://bugzilla.kernel.org/show_bug.cgi?id=5848

sziwan at users.sourceforge.net changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |sziwan at users.sourceforge.net



------- Additional Comments From sziwan at users.sourceforge.net  2006-02-18 14:20 -------
Might this be a serial port configuration issue?  
Back at my last workplace we tested a couple of those (and similar 8250_pci)  
cards and the general results were rather disappointing.  
  
Those cards come with a 16550A that Linux usually configures for 115200, but  
most of them can (and do) handle much more than that. I seem to recall that  
the problem looked like a buffer overrun -- we would get the cards to run at ~  
20 kB/s UMTS download for a couple of seconds, then stall, and sometimes get  
back to a fluctuating 5-15 kB/s, and sometimes not.  
  
When only GPRS was used, the card was much more stable (sustained 4-5 kB/s),  
though occasional hiccups occured.  
  
We initially suspected hardware, but we got to run the cards on the same  
machine with Windows XP, and voila, 33-36 kB/s sustained download rate.  
It might be that we simply couldn't configure the cards properly (if there  
really is _the right way_ to configure serial ports faster than 115200 it  
*really* needs proper docs), but my experience with similar hardware (8250_pci  
GPRS/UMTS modems, serial_cs 3Com Bluetooth PC Card that advertises a 16C950)  
seems to suggest that Linux's support for fast serial ports is just lousy and  
you will run into problems if you can't limit the rate properly (read: GPRS). 
 
Okay, enough ranting, oxiroxt, is the problem less visible if you just use 
GPRS and no UMTS? Checking with recent 2.6.16-rc kernels might be worth 
trying, as some patches were merged that might be relevant to this issue. 

------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.