On 10/29/05, Hui Cheng <hcheng@xxxxxxxxxxx> wrote: > > Thanks for your reply, Fawad :) I have the similar thought as you. I do > believe that the current process would be put to sleep ( like in a wait > queue) at sometime when the request is queued. However, I just cannot find > the code that explicitly do this. I followed all the code from sys_read > until __make_request, at where the request is queued in the request queue. > I was wondering I might miss something. Could anybody help me out of it? > > After your reply, I did some code study from __make_request (ll_rw_blk.c) to actual storage driver and for this I studied the code of drivers/block/floppy.c because floppy.c is the simplest real storage driver. And I found that the sleeping is done by the real hardware driver (which I was thinking earlier too, but now confirm) as when the request_function (do_fd_request) specified by the floppy.c is called, after some processing it calls the process_fd_request which calls the schedule_bh function (defined in floppy.c) schedules the floppy_work with the handler redo_fd_request. When the redo_fd_request is called from the schedule_work, it use make_raw_rw_request to actually fullfil the request and later redo_fd_request calls request_done which through floppy_end_request function ends/signals the kernel that request is done. I hope this will help! You can also work-out on floppy.c in detail ... -- Fawad Lateef -- Kernelnewbies: Help each other learn about the Linux kernel. Archive: http://mail.nl.linux.org/kernelnewbies/ FAQ: http://kernelnewbies.org/faq/
