Matthias Fuchs wrote:
> On Monday 11 January 2010 21:10, Wolfgang Grandegger wrote:
>> We are close, just a few more issues.
> Uff :-)
>
>>> +struct header_msg {
>>> + __u8 len; /* len is always the total message length in 32bit words */
>>> + __u8 cmd;
>>> + __u8 rsvd[2];
>>> +};
>> Please use just one type set. Either u8 or __u8 but not both. The same
>> for __u32, __le32, etc.
> Good point. Marc invented some versions before to use __le32 instead of
> u32 in these device structures.
> As far as I understand this should signal that these are some kind
> of exchange types (typically with userspace programs or in this case
> a device).
Exchange with a device?
> I am using __ types for all structures that are exchanged with the hardware.
> All other datatypes stay in the kernel.
>
> So what's correct?
My understanding is that the types prefixed with "__" are for header
files shared with user space. For the kernel we should use u8 or uint8_t
and friends. Obviously, there is no le32 type - don't asked me why -
therefore just keep __le32. See for example:
http://lxr.linux.no/#linux+v2.6.32/drivers/scsi/aacraid/aacraid.h
[snip]
>>> + * The IDADD message takes up to ESD_MAX_ID_SEGMENT 32 bit
>>> + * bitmasks = 2048 bits. A cleared bit disables reception
>>> + * of the corresponding CAN identifier.
>>> + * The next bitmask value following the CAN 2.0A
>>> + * bits is used to disable reception of extended CAN frames
>>> + * at all.
>>> + */
>>> + msg.msg.hdr.cmd = CMD_IDADD;
>>> + msg.msg.hdr.len = 1 + ESD_MAX_ID_SEGMENT + 1;
>> msg.msg.hdr.len = ESD_MAX_ID_SEGMENT + 2;
>>
>> Please.
> Ok, this was for readability: 1 (msg header) + ESD_MAX_.... (for 2.0a IDs) and
> + 1 (for 2.0b ID range). But you are rigth it looks a little bit curious.
Ah, that was really obvious ;-).
> I will change it and add some more comments about how the device's ID filter
> is configured.
Good.
Wolfgang.
_______________________________________________
Socketcan-core mailing list
[email protected]
https://lists.berlios.de/mailman/listinfo/socketcan-core
