- 08 Mar, 2012 40 commits
-
-
Jiri Slaby authored
Instead of digging a tty out of the tty_driver struct, which is not defined to work, use tty_port properly. This includes proper tty refcounting even though there is no possible race currently. But we will need tty_port for tty buffers in the future anyway. Signed-off-by:
Jiri Slaby <jslaby@suse.cz> Cc: Kyle McMartin <kyle@mcmartin.ca> Cc: Helge Deller <deller@gmx.de> Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Signed-off-by:
Greg Kroah-Hartman <gregkh@linuxfoundation.org>
-
Jiri Slaby authored
Assign the pointer to pdc_console_tty_driver (a tty_driver) earlier. Otherwise the timer may dereference NULL. Signed-off-by:
-
Jiri Slaby authored
The timer is initialized too late. tty->open may fire an invalid timer. So initialize the timer earlier using DEFINE_TIMER. Signed-off-by:
-
Jiri Slaby authored
The tty->count test in the timer was racy. Let's remove the test and properly delete the timer and wait for the body to finish using _sync version of del_timer. Signed-off-by:
-
Jiri Slaby authored
Again, no need to duplicate the code. Let's use the helper. Amiserial changes are only free of compilation errors. I have no access to the hardware. Signed-off-by:
-
Jiri Slaby authored
Hmm, the code was sleeping with interrupts disabled. This was not good. Fix this by turning interrupts at an appropriate place. (The race is protected by CLOSING flag.) After the move, the code is identical to tty_port_close_end, so use it! Signed-off-by:
-
Jiri Slaby authored
Hmm, 150 lines of duplicated stuff is gone now. Signed-off-by:
-
Jiri Slaby authored
This is a preparation for a switch to tty_port_block_til_ready. We need amiga_carrier_raised and amiga_dtr_rts. The implementation is taken from startup, shutdown and current block_til_ready. Signed-off-by:
-
Jiri Slaby authored
amiserial is the last user of serialP.h. Let's move struct serial_state directly to amiserial and remove serialP crap from includes. Finally, remove the header from the tree completely. Signed-off-by:
-
Jiri Slaby authored
* instead of line, use tty->index or iterator... * irq and type are left unset. So get rid of them. Signed-off-by:
-
Jiri Slaby authored
And use it to make the code more readable. Signed-off-by:
-
Jiri Slaby authored
* remove pointless checks (tty cannot be NULL at that points) * fix some printks (use __func__, print text directly w/o using global strings) * remove some empty lines This is the last patch for simserial. Overall, the driver is 400 lines shorter. Being now at 560 lines. It was tested using ski with a busybox userspace. Signed-off-by:
-
Jiri Slaby authored
Make the code to conform to the standard. Also make it readable. Signed-off-by:
-
Jiri Slaby authored
Use headers from linux/* instead of asm/. Remove declaration of console_drivers, it's in linux/console.h already. Signed-off-by:
-
Jiri Slaby authored
Or the obsolete ones like: "Let's have a little bit of fun" I have never had fun with software. For fun, one needs hard-ware. Signed-off-by:
-
Jiri Slaby authored
Convert shutdown to be tty_port_operations->shutdown. Then we can use tty_port_hangup. (And we have to use tty_port_close.) This means we no longer touch ASYNC_INITIALIZED, TTY_IO_ERROR. Also we do not need to do any peculiar TTY logic in the file now. Signed-off-by:
-
Jiri Slaby authored
So now we convert startup to be ->activate of tty_port. This means we no longer care about INITIALIZED and TTY_IO_ERROR flags. After we have ->activate much of the code may go as it duplicates what tty_port_open does. In this case tty_port_open adds block_til_ready to the path. But we do not define carrier hooks, so it is a noop. Signed-off-by:
-
Jiri Slaby authored
So that we will not be surprised in the ISR anymore. Signed-off-by:
-
Jiri Slaby authored
I.e. remove more copied bloat. The only change is that we wait_until_sent now. Which is what we really should do. Signed-off-by:
-
Jiri Slaby authored
The code is identical except locking. But added locks to protect counts do not hurt here. Rather the contrary. Signed-off-by:
-
Jiri Slaby authored
All ->start, ->stop and ->wait_until_sent are empty and need not be defined. The time to remove them is now. Signed-off-by:
-
Jiri Slaby authored
And use it to make the code more readable. Since tport doesn't conflict with port anymore and there are not many tport accessors left, do also s/\<tport\>/port/g. Signed-off-by:
-
Jiri Slaby authored
Let's do a spin-off of serial_state structure with only needed elements. And remove serialP crap from includes. Signed-off-by:
-
Jiri Slaby authored
* instead of line, use tty->index or an iterator * icount is not made public, only the tx path increments it Signed-off-by:
-
Jiri Slaby authored
It is totally unused. Signed-off-by:
-
Jiri Slaby authored
We do not use any of the preinitialized rs_state members for something real. So there is no need to initialize them. At the places we used them for printing, just print the values. And since only one port is supported, get rid of the loop. This simplifies simrs_init a heap. Thus we can handle fail paths in a standard way without panicing. Signed-off-by:
-
Jiri Slaby authored
This changes flags' type to ulong which is appropriate for all the set/clear_bits performed in the drivers.. Signed-off-by:
-
Jiri Slaby authored
Nothing special. Just remove count from serial_state and change all users to use tty_port. Signed-off-by:
-
Jiri Slaby authored
Note that previously simserial set the delay to 0. So we preserve that. BUT, is it correct? Signed-off-by:
-
Jiri Slaby authored
Add tty_port to serial_state and start using common tty port members from tty_port in amiserial and simserial. The rest will follow one by one. Signed-off-by:
-
Jiri Slaby authored
This avoids pain with tty refcounting and touching tty_port in the future. It allows us to remove some state->tty tests because the tty passed down to them can never be NULL. Signed-off-by:
-
Jiri Slaby authored
This avoids pain with tty refcounting and touching tty_port in the future. It allows us to remove some info->tty tests because the tty passed down to them can never be NULL. Signed-off-by:
-
Jiri Slaby authored
Do not copy whole serial_state. We only need to know whether the speed is to be changed. Hence store the info in advance and use it later. A simple bool is enough. Also remove reduntant assignments and move the tests directly to the 'if'. Signed-off-by:
-
Jiri Slaby authored
This is the final step to get rid of the one of the structures. A further cleanup will follow. And I struct serial_state deserves cease to exist after a switch to tty_port too. While changing the lines, it removes also pointless tty->driver_data casts. Signed-off-by:
-
Jiri Slaby authored
They used to work as a storage for 'info' pointer used in ISRs. They are not really needed. Just pass the pointer through request_irq to the handlers. It was set to NULL and tested in the ISRs, but we do not need the tests as we disable all the interrupts at the same places where NULL sets were. Signed-off-by:
-
Jiri Slaby authored
We do not set ASYNC_SHARE_IRQ anywhere. And since IRQF_DISABLED is a noop, pass zero to request_irq directly instead of this ugly macro. Signed-off-by:
-
Jiri Slaby authored
It never worked there. The ISR was never written for that kind of stuff. So remove all that crap with a hash of linked lists and pass the pointer directly to the ISR. BTW this answers the question there: * I don't know exactly why they don't use the dev_id opaque data * pointer instead of this extra lookup table -> Because they thought they will support more devices bound to a single interrupt w/o IRQF_SHARED. They would need exactly the hash there. What I don't understand is rebinding of the interrupt in the shutdown path. They perhaps meant to do just synchronize_irq? In any case, this is all gone and free_irq there properly. By removing the hash we save some bits (exactly NR_IRQS * 8 bytes of .bss and over a kilo of .text): before: text data bss dec hex filename 19600 320 8227 28147 6df3 ../a/ia64/arch/ia64/hp/sim/simserial.o after: text data bss dec hex filename 18568 320 28 18916 49e4 ../a/ia64/arch/ia64/hp/sim/simserial.o Note that a shared interrupt could not work too. request_irq requires data parameter to be non-NULL. So the whole IRQ_T exercise was pointless. Finally, this helps us remove another two members of async_struct :). Signed-off-by:
-
Jiri Slaby authored
This means: * close_delay * closing_wait * line * port * xmit_fifo_size This actually fixes a bug in amiserial. It initializes one and uses the other of the close delays. Yes, duplicating structure members is evil. Signed-off-by:
-
Jiri Slaby authored
The same as for amiserial. Use only one instance of the flags. Also remove them from async_struct now. Nobody else uses them. Signed-off-by:
-
Jiri Slaby authored
Without this, the code succeeds when the port is opened by root and we get unwanted interrupts storm on the first key stroke. Instead of that, tell the user we failed and that we won't continue. I suppose, the code was copied from the serial layer where we may want to change the irq number, so we must allow open even of the failing port. This is not the case for this driver at all. Signed-off-by:
Tony Luck <tony.luck@intel.com> Signed-off-by:
-
