On Fri 2015-07-17 10:51:11, atull@xxxxxxxxxxxxxxxxxxxxx wrote:
> From: Alan Tull <atull@xxxxxxxxxxxxxxxxxxxxx>
>
> Add a document on the new FPGA manager core.
>
> --- /dev/null
> +++ b/drivers/staging/fpga/Documentation/fpga-mgr.txt
> @@ -0,0 +1,117 @@
> + FPGA Manager Core
> +
> + Alan Tull 2015
> +
> + Overview
> + --------
The formatting is quite funny here. Add a newline after ---'s? Or
better format it the way other documents are formatted?
> +The FPGA manager core exports a set of functions for programming an image to a
"into"?
> +FPGA. All manufacturor specifics are hidden away in a low level driver. The
manufacturer (more then one instance).
> +API is manufacturor agnostic. Of course the FPGA image data itself is very
> +manufacturor specific but for our purposes it's just data in a buffer or file
, but
> +or something. The FPGA manager core won't parse it or know anything about it.
kill "or know anything"?
> + Files
> + -----
> +drivers/staging/fpga/fpga-mgr.c
> +include/linux/fpga/fpga-mgr.h
> +
Kill this section, as it is going to change?
> + The API Functions
> + ----------------
> +The API that is exported is currently 6 functions:
> +
> + int fpga_mgr_buf_load(struct fpga_manager *mgr,
> + u32 flags,
> + const char *buf,
> + size_t count);
> +
> +An FPGA image exists as a buffer in memory. Load it into the FPGA. The FPGA
> +ends up in operating mode or return a negative error code.
So 0 on success?
> + int fpga_mgr_firmware_load(struct fpga_manager *mgr,
> + u32 flags,
> + const char *image_name);
> +
> +An FPGA image exists as a file that is on the firmware search path (see the
", that is in"
> +firmware class documentation). Load as above.
> +
> + struct fpga_manager *of_fpga_mgr_get(struct device_node *node);
> +
> +Given a DT node, get a reference to a fpga manager.
fpga->FPGA, fix globally??
> + void fpga_mgr_put(struct fpga_manager *mgr);
> +
> +Release the reference to the fpga manager.
> +
> + int fpga_mgr_register(struct device *dev,
> + const char *name,
> + const struct fpga_manager_ops *mops,
> + void *priv);
> + void fpga_mgr_unregister(struct device *dev);
> +
> +Register/unregister the lower level device specific driver.
"low level".. And "device specific" -> "FPGA-specific" ?
> +To add another fpga manager, look at the bottom part of socfpga.c for an
> +example, starting with the declaration of socfpga_fpga_ops.
Umm. You have good documentation below. Maybe you don't need to send
people to read sources...?
> +static const struct fpga_manager_ops socfpga_fpga_ops = {
> + .write_init = socfpga_fpga_ops_configure_init,
> + .write = socfpga_fpga_ops_configure_write,
> + .write_complete = socfpga_fpga_ops_configure_complete,
> + .state = socfpga_fpga_ops_state,
> +};
> +
> +You will want to create a platform driver that has a set of ops like that
> +and then register it with fpga_mgr_register in your probe function. Your
> +ops will implement whatever device specific register writes needed and
> +will return negative error codes if things don't go well.
> +
> +The programming seqence is:
sequence.
> + 1. .write_init
> + 2. .write (may be called once or multiple times)
> + 3. .write_complete
> +
> +The .write_init function will prepare the FPGA to receive the image data.
> +
> +The .write function receives an image buffer or a chunk of the image and
> +writes it the FPGA. The buffer may arrive as one chunk or a bunck
> of
bunch.
> +small chunks through this function being called multiple times.
> +
> +The .write_complete function is called after all the image has been written
> +to put the FPGA into operating mode.
> +
> +The .state function will read your hardware and return a code of type
> +"enum fpga_mgr_states". It doesn't result in a change in hardware state.
--
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
