search

daniel-santos / mcp2210-linux

MCP2210 driver for linux
http://danielthesantos.blogspot.com/search/label/mcp2210
51 stars 31 forks source link

SPI driver runtime error messages: check_response: mcp2210 command failed (status = 0xf8) (device busy) #26

Open CajusH opened 8 years ago

CajusH commented 8 years ago

Hi,

I am using your MCP2210 driver with my Kernel 4.4.10 on an i.MX6 design. I have two MCP2210 in my design, both chips drive SPI devices. I monitored the SPI communication and found out, that after 60 bytes on the SPI there is a long delay (19ms and more) The attached screenshot shows a transfer of 260 bytes (4 blocks with 60 bytes and one with 20 bytes) scope_0

Do you have an explanation for this and maybe a solution to get rid of the delays? We typically have larger data blocks then 60 bytes.

Depending on the clock I also see error messages:

Transfer 260 bytes spi mode: 0 bits per word: 8 max speed: 100000 Hz (100 KHz)
usb 2-1.2: check_response: mcp2210 command failed (status = 0xf8) usb 2-1.2: report_status_error: status 0xf8: device busy usb 2-1.2: spi_mcp_error: cmd->busy_count 0 usb 2-1.2: complete_urb: ignoring mcp_status error and repeating usb 2-1.2: check_response: mcp2210 command failed (status = 0xf8) usb 2-1.2: report_status_error: status 0xf8: device busy usb 2-1.2: spi_mcp_error: cmd->busy_count 1 usb 2-1.2: complete_urb: ignoring mcp_status error and repeating usb 2-1.2: check_response: mcp2210 command failed (status = 0xf8) usb 2-1.2: report_status_error: status 0xf8: device busy usb 2-1.2: spi_mcp_error: cmd->busy_count 2 usb 2-1.2: complete_urb: ignoring mcp_status error and repeating usb 2-1.2: check_response: mcp2210 command failed (status = 0xf8) usb 2-1.2: report_status_error: status 0xf8: device busy usb 2-1.2: spi_mcp_error: cmd->busy_count 3 usb 2-1.2: complete_urb: ignoring mcp_status error and repeating

It looks like the number of error message correspond to the number transferred bytes and the clock. If I use a higher clock (more than 430kHz) I will not get the errors for my 260 byte transfer, but I will get an error, if I transfer 460 bytes with the 430kHz clock.

Any idea, how to get rid of the error messages?

Best Regards Cajus

daniel-santos commented 5 years ago

I'm really sorry I missed this message. This is because I've written a supreme driver for a shitty device. I have some local commits that better tunes this, but on the order of +20% performance. I'm going to continue maintaining this driver and get those improvements committed once I clean up and test better, but it's just a falsely advertised product!

jhausladen commented 3 years ago

@daniel-santos Were there any updates on this? At least, I still get a lot of those error messages, flooding the kernel logs... spi_mcp_error: cmd->busy_count 0 complete_urb: ignoring mcp_status error and repeating

CajusH commented 3 years ago

I simply commended-out the corresponding lines as the warning seems to appear on every message > 60 bytes long, even if the data was transferred.