2010/10/15 Marc Kleine-Budde: > I've applied the patch to net-next-2.6: > http://git.pengutronix.de/?p=mkl/linux-2.6.git;a=shortlog;h=refs/heads/can/mcp251x-async > Andre, have a look at my branch before you start hacking. > Can I have your S-o-b for the patch?
Thanks! I am sending the new patch (separate email, with signoff), already with the changes you did in your git, and some more things, as suggested. > - implement IFF_ECHO Done. > +static void mcp2515_read_rxb(...) > I suggest to change the function's semantics, replace the instruction > parameter by the buffer number remove the complete parameter. Do the > demultiplexing of buffer to instruction and complete function here. Didn't do this yet. I'll wait to see if everyone approves this new mcp2515 driver. If so, I'll then work on these functions, to also support the MCP2510. > + priv->skb = skb; > With the current driver I have the situation, that xmit gets called even > if the skb has not been deliverted yet. I don't know why.... That should never happen, I believe. But let me know if this ever happens with this driver, please. > If the driver performs better I'd like to replace the existing driver. Ok, but I'm biased. I'll leave it for you all to decide. :-) I believe that using the spi_async API directly is the way to go here. In the tests I did, the overhead of the mcp251x workqueues were not negligible (between 50us and 100us, that's about the time it takes to receive one entire frame at 1Mbps, and thus nearly approaching buffer overrun), as you can see in the oscilloscope screenshots here: http://clientes.netvisao.pt/anbadeol/can.html I noticed that the mcp251x driver has been improved since, so I don't know if the performance difference between the two is still that big. I'll try to reassemble my test set up (Gumstix Verdex PXA270 400MHz and MCP2515) and grab some new oscilloscope shots comparing with the current mcp251x, soon... 2010/10/12 David Jander: > Can't wait to try it out! > Looking at the source, it really looks good. Be certain that I'll test it > extensively once you re-post the modified version... Thanks! -- Andre B. Oliveira http://clientes.netvisao.pt/anbadeol/ _______________________________________________ Socketcan-core mailing list [email protected] https://lists.berlios.de/mailman/listinfo/socketcan-core
