src - OpenBSD base system

Age	Commit message (Collapse)	Author
2010-01-07	Repair non-MULTIPROCESSOR kernel compilation.	Miod Vallat

2010-01-07	Make sure to update p_cpu on MULTIPROCESSOR kernels.	Miod Vallat

2010-01-07	Rename _rxfilter functions to _iff for consistency. From Brad, ok kevlo@	Stuart Henderson

2010-01-07	dont use the xs in the scsi_cmd routines after calling ciss_cmd, it might	David Gwynne
	have been completed. the xs was used to figure out if we should return COMPLETED or SUCCESFULLY_QUEUED. the return value doesnt mean anything anymore, so just return COMPLETE.
2010-01-07	CISS_UNLOCK before returning from the scsi_cmd handler when we hit NO_CCB.	David Gwynne
	this should solve problems like this: WARNING: SPL NOT LOWERED ON SYSCALL 16384 8 EXIT dae4560 6 Stopped at Xsyscall+0x1eb: movl $0,%gs:0x3d4 ddb{3}>
2010-01-07	Enable IPI interrrupt in trap. ok miod@	Takuya ASADA

2010-01-07	warn if scsi_done has been called twice on an xs in the scsi_xs_sync path	David Gwynne
	rather than panic. ok krw@
2010-01-06	sync	Stuart Henderson

2010-01-06	Add Canon PowerShot SX100, Linksys WUSB600N v2 (new version of RT2870). From ↵	Stuart Henderson
	Brad.
2010-01-06	sync	Stuart Henderson

2010-01-06	New Broadcom nics, from Linux driver via Brad.	Stuart Henderson

2010-01-05	- properly spell "packet".	Jasper Lievisse Adriaanse

2010-01-05	dont need to end statements with two semicolons. one is enough.	David Gwynne
	pointed out by miod@ and claudio@
2010-01-05	Dynamic allocation for ASID and ASID generation number on struct pmap. ok miod@	Takuya ASADA

2010-01-05	only call disk_unbusy when we're finished with the io (or we're requeuing	David Gwynne
	it when we hit NO_CCB). retried io would call unbusy twice before this change. noticed by deraadt@
2010-01-05	atapiscsi didnt call scsi_done when the xfer had SCSI_POLL set. this is	David Gwynne
	Bad. this sort of worked before because the midlayer would sleep on the ITSDONE flag in the scsi_xfer, which atapiscsi does set. however, with the introduction of scsi_xs_sync, it now sleeps on a condition that is only cleared if the adapter properly and correctly calls scsi_done when it is in fact done with the xfer. sorry guys. ok krw@
2010-01-04	Previous commit was too restrictive: just ignore the CLONE_FS flag	Philip Guenthe
	when not doing CLONE_THREAD, like we did before, instead of treating it as an error. This unbreaks Opera running in linux compat. Also, Linux completely ignores CLONE_DETACHED, so do that too, and consistently refer to the flags in comment via their Linux names without trying to abbreviate them further. Problem found and fix confirmed by sturm@
2010-01-04	remove dead assignment and newly created unused variable.	Charles Longeau
	Found by LLVM/Clang Static Analyzer. ok krw@ dlg@
2010-01-04	turn off verbose	Theo de Raadt

2010-01-04	nothing in the midlayer cares about ITSDONE anymore. if a command is done,	David Gwynne
	the adapter will call scsi_done on it.
2010-01-04	rename ccb_xs to ccb_cookie and change its type from a scsi_xfer * to a	David Gwynne
	void *. this lets us pass things other than scsi_xfers to ccb completion handlers.
2010-01-04	protect the ccb free queue with a mutex rather than splbio	David Gwynne

2010-01-04	Don't decrement the refcnt on a plimits until after we're done	Philip Guenthe
	copying it, so that the process can't sleep in pool_get() and have the source structure get pool_put() or modified behind its back. ok deraadt@
2010-01-04	Add a few missing scsi_done()'s before returning COMPLETE. Also	Kenneth R Westerback
	return COMPLETE and not CMD_COMPLETE in one case. Tested by kettenis@ and myself on Blade1000's. ok dlg@
2010-01-04	introduce scsi_xs_sync. this will synchronously execute a scsi_xfer	David Gwynne
	and do the appropriate sleeps, retries, error processing, and finally returns an errno style value to the caller. this cuts scsi_scsi_cmd, the ioctl code, sd_flush, and scsi_inquiry over to scsi_xs_sync. ok krw@
2010-01-03	Implement pmap_proc_iflush() such that the instruction cache is synchronized	Mark Kettenis
	with the data cache when ptrace(2) is used to write into a process' address space. ok miod@
2010-01-03	Use atomic operations to access the per-cpu scheduler flags.	Mark Kettenis

2010-01-03	delete unused variable	Theo de Raadt

2010-01-03	enable aibs	Theo de Raadt

2010-01-03	change M_WAITOK --> M_WAITOK\|M_CANFAIL	Charles Longeau
	with input from marco@ ok marco@ krw@ this was ok looooooooong time ago but lost in my mailbox
2010-01-03	Use a split genassym.cf scheme on mips64 ports. Don't forget to rerun config(8).	Miod Vallat

2010-01-03	disk cache mode page	David Gwynne

2010-01-03	oops, get the order of args right for the header request in	David Gwynne
	mpi_bio_get_pg0_raid. the sensor updates dont poll at all now.
2010-01-03	mpi_bio_get_pg0_raid is only called from a process context. let the cfg	David Gwynne
	requests sleep rather than poll for completion.
2010-01-03	oops, forgot to set the ccb_cookie in mpi_wait. mpi_wait_done would fault	David Gwynne
	when it used the uninitialized cookie.
2010-01-03	get rid of the last internal user of splbio. waiting for the completion of	David Gwynne
	a ccb can now be done with mpi_wait. this switches the cfg page handlers over from their own tsleep stuff to mpi_wait.
2010-01-03	rename ccb_xs to ccb_cookie, and switch it from a struct scsi_xfer * to a	David Gwynne
	void *. this will let me stash things other than scsi xfers in the ccb for ccb_done handlers to use.
2010-01-03	dont leak a ccb if we fail to get a reply in portenable	David Gwynne

2010-01-03	when getting a reply from the hw, only sync the dmamem for that one reply	David Gwynne
	rather than all the replies.
2010-01-03	switch mpi from using splbio to protect itself over to mutexes.	David Gwynne
	mpi only needs two mutexes, one for the list of free ccbs, and another to protect the reply doorbell. the latter is necessary to allow polling for command completion to work in smp systems. tested on sas and fc hbas. this diff was written over 2 years ago now with surprisingly few tweaks to handle changes that have occurred since then.
2010-01-03	scsi_done sets ITSDONE, we dont have to do it.	David Gwynne

2010-01-02	move the buf handling from using scsi_scsi_cmd over to scsi_xs_exec. i	David Gwynne
	think this is the last chunk of code using a buf pointer in the scsi_xfer now. this has not been tested due to a lack of hardware anywhere. if there are problems please report them as soon as possible. ok krw@ miod@
2010-01-02	uninitalized protocol version for ipv6; from mickey; ok claudio	Markus Friedl

2010-01-02	When finding certain novatel devices in mass storage mode	Jonathan Gray
	we need to do scsi ejects. One reported/tested by lahrcm@hotmail.com another suggested by David Coppa based on what FreeBSD does.
2010-01-02	We need an additional level of indirection through 'struct scsibus_attach_args'	Mark Kettenis
	to find the 'struct scsi_link' associated with a scsibus(4). Interpreting 'struct scsibus_attach_args' as a 'struct scsi_link' happened to mostly work since it had a zero stored in the right place. However, after dlg@'s changes to 'struct scsi_link' that no longer turned out to be true. Makes my blade1k find its root disk again.
2010-01-01	Miod must have gotten his fingers stuck between the keys on his keyboard and	Mark Kettenis
	accidentally swapped to lines of code. Swap them back such that this compiles again.
2010-01-01	Map the device registers before using them. We were lucky that some other	Mark Kettenis
	device already mapped the block that contains our registers, but if we change the probe order, or disable devices we might not end up so lucky. ok jsing@, miod@
2010-01-01	Make sure than ptrace functions never return a non-zero value in the zero	Miod Vallat
	register.
2010-01-01	Make sure we grab the kernel lock before invoking trapsignal().	Miod Vallat

2010-01-01	If you want to use atomic ops, you need to #include the proper files instead	Miod Vallat
	of relying upon other headers bringing it in for you.