Re: [RFC PATCH v2 2/6] i2c: allow DT nodes without 'compatible'

[Kieran Bingham]

Hi Wolfram,

On 15/04/2020 09:16, Wolfram Sang wrote:
> 
>> Aha, is it easy enough to distinguish that difference in user-space so
>> that we can present a specific character to indicate this in i2cdetect?
>> Or is that not so easy?
> 
> I thought about it shortly but have not come up with a way of doing
> that. This is the code in i2cdetect:
> 
> 	/* Set slave address */
> 	if (ioctl(file, I2C_SLAVE, i+j) < 0) {
> 		if (errno == EBUSY) {
> 			printf("UU ");
> 			continue;
> 		} else {
> 			fprintf(stderr, "Error: Could not set "
> 				"address to 0x%02x: %s\n", i+j,
> 				strerror(errno));
> 			return -1;
> 		}
> 	}
> 
> So, if we chose to use another errno to indicate 'reserved' and update
> i2cdetect, all old versions of i2cdetect will have ugly error messages.
> And adding another IOCTL just for printing reserved addresses neither
> sounds great.

Indeed, a different errno would be about all we could do - and it would
seemingly leave old i2cdetects with complete failures, if it goes
through that non-EBUSY code path.

Ayeeeee :-S

--
Kieran