Mostly some maintainence patches. I'm happy to take feedback. Patch 2 is useful for the mtp function, since it uses endpoint size to figure out when to read/write a zero length packet. It would be nice to get these values at the start of a connection and cache them for the duration, although it's not a must-have.
I'd also like some direction on another patch I'm planning: Android has migrated to using f_fs for the mtp function, but in the process we ran into some performance hiccups in comparison to f_mtp. Allocating memory at boot allowed f_mtp the use of very large buffers, around 250kB. Compared to using a 16 or 32 kB buffer, the larger buffer is faster by around 50%. Functionfs can't use io sizes that large due to memory fragmentation. One option here would be to add O_DIRECT support. Then on each io, we could get_user_pages, vmap them together, and pass that address into ep_queue. This would would allow any size to be used and eliminates a copy, but would be rather involved. Does this sound like something that people want or find acceptable? If there's a better option, I'd be willing to help with that as well. Jerry Zhang (3): usb: gadget: function: f_fs: Fix doc for FUNCTIONFS_INTERFACE_REVMAP usb: gadget: function: f_fs: Let ffs_epfile_ioctl wait for enable. usb: gadget: function: f_fs: Move epfile waitqueue to ffs_data. drivers/usb/gadget/function/f_fs.c | 103 ++++++++++++++++++++---------------- drivers/usb/gadget/function/u_fs.h | 3 ++ include/uapi/linux/usb/functionfs.h | 8 ++- 3 files changed, 62 insertions(+), 52 deletions(-) -- 2.12.2.816.g2cccc81164-goog -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
