Jean Delvare [Thu, 11 Aug 2005 21:51:10 +0000 (23:51 +0200)]
[PATCH] I2C: Kill i2c_algorithm.id (6/7)
In theory, there should be no more users of I2C_ALGO_* at this point.
However, it happens that several drivers were using I2C_ALGO_* for
adapter ids, so we need to correct these before we can get rid of all
the I2C_ALGO_* definitions.
Note that this also fixes a bug in media/video/tvaudio.c:
/* don't attach on saa7146 based cards,
because dedicated drivers are used */
if ((adap->id & I2C_ALGO_SAA7146))
return 0;
This test was plain broken, as it would succeed for many more adapters
than just the saa7146: any those id would share at least one bit with
the saa7146 id. We are really lucky that the few other adapters we want
this driver to work with did not fulfill that condition.
Signed-off-by: Jean Delvare <khali@linux-fr.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Jean Delvare [Thu, 11 Aug 2005 21:41:56 +0000 (23:41 +0200)]
[PATCH] I2C: Kill i2c_algorithm.id (5/7)
Merge the algorithm id part (16 upper bits) of the i2c adapters ids
into the definition of the adapters ids directly. After that, we don't
need to OR both ids together for each i2c_adapter structure.
Signed-off-by: Jean Delvare <khali@linux-fr.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Jean Delvare [Thu, 11 Aug 2005 21:38:52 +0000 (23:38 +0200)]
[PATCH] I2C: Kill i2c_algorithm.id (3/7)
Don't rely on i2c_algorithm.id to alter the i2c adapter's id, use the
I2C_ALGO_* value directly instead, because i2c_algorithm will soon
have no id member no more.
Signed-off-by: Jean Delvare <khali@linux-fr.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Jean Delvare [Thu, 11 Aug 2005 21:36:49 +0000 (23:36 +0200)]
[PATCH] I2C: Kill i2c_algorithm.id (2/7)
Use the adapter id rather than the algorithm id to detect the i2c-isa
pseudo-adapter. This saves one level of dereferencing, and the
algorithm ids will soon be gone anyway.
Signed-off-by: Jean Delvare <khali@linux-fr.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Jean Delvare [Thu, 11 Aug 2005 21:33:24 +0000 (23:33 +0200)]
[PATCH] I2C: Kill i2c_algorithm.name (1/7)
The name member of the i2c_algorithm is never used, although all
drivers conscientiously fill it. We can drop it completely, this
structure doesn't need to have a name.
Signed-off-by: Jean Delvare <khali@linux-fr.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The 24RF08 corruption would better be prevented at i2c-core level than
at chip driver level, for several reasons:
* The second quick write should happen as soon as possible after the
first one, so as to limit the risk that another command is issued on
the bus inbetween, causing the corruption.
* As a matter of fact, the protection code at driver level was reworked
at least three times already, which proves how hard it is to get it
right there, while it's straightforward at i2c-core level.
* It's easy to add a new driver that would need the protection, and
forget to add it. This did happen already.
* As additional probing addresses can be passed to most i2c chip drivers
as module parameters, virtually every i2c chip driver would need the
protection if we want to be really safe.
* Why duplicate code when we can easily avoid it?
Signed-off-by: Jean Delvare <khali@linux-fr.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Jean Delvare [Tue, 9 Aug 2005 18:17:55 +0000 (20:17 +0200)]
[PATCH] I2C: Rewrite i2c_probe
i2c_probe was quite complex and slow, so I rewrote it in a more
efficient and hopefully clearer way.
Note that this slightly changes the way the module parameters are
handled. This shouldn't change anything for the most common cases
though.
For one thing, the function now respects the order of the parameters
for address probing. It used to always do lower addresses first. The
new approach gives the user more control.
For another, ignore addresses don't overrule probe addresses anymore.
This could have been restored the way it was at the cost of a few more
lines of code, but I don't think it's worth it. Both lists are given
as module parameters, so a user would be quite silly to specify the
same addresses in both lists. The normal addresses list is the only
one that isn't controlled by a module parameter, thus is the only one
the user may reasonably want to remove an address from.
Another significant change is the fact that i2c_probe() will no more
stop when a detection function returns -ENODEV. Just because a driver
found a chip it doesn't support isn't a valid reason to stop all
probings for this one driver. This closes the long standing lm_sensors
ticket #1807.
In terms of algorithmic complexity, the new code is way better. If
I is the ignore address count, P the probe address count, N the
normal address count and F the force address count, the old code
was doing 128 * (F + I + P + N) iterations max, while the new code
does F + P + ((I+1) * N) iterations max. For the most common case
where F, I and P are empty, this is down from 128 * N to N.
Signed-off-by: Jean Delvare <khali@linux-fr.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Jean Delvare [Mon, 1 Aug 2005 20:50:08 +0000 (22:50 +0200)]
[PATCH] hwmon: hwmon vs i2c, second round (10/11)
I see very little reason why vid_from_reg is inlined. It is not
exactly short, its parameters are seldom known in advance, and it is
never called in speed critical areas. Uninlining it should cause
little performance loss if any, and saves a signficant space as well
as compilation time.
As suggested by Alexey Dobriyan, I am leaving vid_to_reg inline for now,
as it is short and has a single user so far.
Signed-off-by: Jean Delvare <khali@linux-fr.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Jean Delvare [Sun, 31 Jul 2005 19:52:01 +0000 (21:52 +0200)]
[PATCH] hwmon: hwmon vs i2c, second round (07/11)
The only part left in i2c-sensor is the VRM/VRD/VID handling code.
This is in no way related to i2c, so it doesn't belong there. Move
the code to hwmon, where it belongs.
Note that not all hardware monitoring drivers do VRM/VRD/VID
operations, so less drivers depend on hwmon-vid than there were
depending on i2c-sensor.
Signed-off-by: Jean Delvare <khali@linux-fr.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Jean Delvare [Sun, 31 Jul 2005 19:49:03 +0000 (21:49 +0200)]
[PATCH] hwmon: hwmon vs i2c, second round (06/11)
The only thing left in i2c-sensor.h are module parameter definition
macros. It's only an extension of what i2c.h offers, and this extension
is not sensors-specific. As a matter of fact, a few non-sensors drivers
use them. So we better merge them in i2c.h, and get rid of i2c-sensor.h
altogether.
Signed-off-by: Jean Delvare <khali@linux-fr> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Jean Delvare [Sun, 31 Jul 2005 19:36:24 +0000 (21:36 +0200)]
[PATCH] hwmon: hwmon vs i2c, second round (03/11)
We now have two identical structures, i2c_address_data in i2c-sensor.h
and i2c_client_address_data in i2c.h. We can kill one of them, I choose
to keep the one in i2c.h as it makes more sense (this structure is not
specific to sensors.)
Signed-off-by: Jean Delvare <khali@linux-fr.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Jean Delvare [Sun, 31 Jul 2005 19:33:23 +0000 (21:33 +0200)]
[PATCH] hwmon: hwmon vs i2c, second round (02/11)
The way i2c-sensor handles forced addresses could be optimized. It
defines a structure (i2c_force_data) to associate a module parameter
with a given kind value, but in fact this kind value is always the
index of the structure in each array it is used in. So this additional
value can be omitted, and still be deduced in the code handling these
arrays.
Signed-off-by: Jean Delvare <khali@linux-fr.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Jean Delvare [Sun, 31 Jul 2005 19:20:43 +0000 (21:20 +0200)]
[PATCH] hwmon: hwmon vs i2c, second round (01/11)
Add support for kind-forced addresses to i2c_probe, like i2c_detect
has for (essentially) hardware monitoring drivers.
Note that this change will slightly increase the size of the drivers
using I2C_CLIENT_INSMOD, with no immediate benefit. This is a
requirement if we want to merge i2c_probe and i2c_detect though, and
seems a reasonable price to pay in comparison with the previous
cleanups which saved much more than that (such as the i2c-isa cleanup
or the i2c address ranges removal.)
Signed-off-by: Jean Delvare <khali@linux-fr.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Jean Delvare [Sun, 31 Jul 2005 17:02:53 +0000 (19:02 +0200)]
[PATCH] I2C: Improve core debugging messages
The debugging messages in i2c-core are more confusing than helpful. Some
lack their trailing newline, some lack a prefix, some are redundant,
some lack precious information. Here is my attempt to introduce some
standardization in there.
I also changed two messages in i2c-dev to make it clear they come from
i2c-dev.
Signed-off-by: Jean Delvare <khali@linux-fr.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Ian Campbell [Wed, 10 Aug 2005 07:51:16 +0000 (08:51 +0100)]
[PATCH] I2C: i2c-algo-pca -- gracefully handle a busy bus
I've been running with this patch for a while now, and while I've never
seen it trigger except with buggy hardware I think it is a cleaner way
to handle a busy bus. I had -EBUSY until about 10 minutes ago but -EIO
seems to be what most of the existing algo drivers will return in the
same circumstances.
Signed-off-by: Ian Campbell <icampbell@arcom.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Jean Delvare [Wed, 27 Jul 2005 19:32:02 +0000 (21:32 +0200)]
[PATCH] hwmon: tag super-i/o find functions __init
Super-I/O find functions in hardware monitoring drivers can be tagged
__init as they are only called from functions themselves tagged __init.
Two of them (smsc47b397 and w83627ehf) already do, but the other four
of them (it87, pc87360, smsc47m1 and w83627hf) did not.
This saves a few bytes of memory after the drivers are loaded, 192 in
the case of the it87 driver.
Signed-off-by: Jean Delvare <khali@linux-fr.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Jean Delvare [Wed, 27 Jul 2005 20:14:49 +0000 (22:14 +0200)]
[PATCH] I2C: refactor message in i2c_detach_client
We could refactor the error message 34 different i2c drivers print if
i2c_detach_client() fails in this function itself. Saves quite a few
lines of code. Documentation is updated to reflect that change.
Note that this patch should be applied after Rudolf Marek's w83792d
patches.
Signed-off-by: Jean Delvare <khali@linux-fr.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
I would like to announce support for W83792D chip. This driver was developed
by Winbond Electronics Corp. I added sysfs attributes callbacks infrastructure
plus various code fixes and codingstyle cleanups. I would like to thank Winbond
for supporting free software.
This patch is against 2.6.13rc3 plus hwmon-class and hwmon-split.
Separate patch for documantation and hwmon class register will follow.
Signed-off-by: Rudolf Marek <r.marek@sh.cvut.cz> Signed-off-by: Chunhao Huang <DZShen@Winbond.com.tw> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Jean Delvare [Tue, 19 Jul 2005 22:05:33 +0000 (00:05 +0200)]
[PATCH] I2C: Separate non-i2c hwmon drivers from i2c-core (8/9)
Kill all uses of i2c_is_isa_adapter except for the hybrid drivers (it87,
lm78, w83781d). The i2c-isa adapter not being registered with the i2c
core anymore, drivers don't have to fear being erroneously attached to
it.
Jean Delvare [Tue, 19 Jul 2005 22:02:32 +0000 (00:02 +0200)]
[PATCH] I2C: Separate non-i2c hwmon drivers from i2c-core (7/9)
Kill normal_isa in header files, documentation and all chip drivers, as
it is no more used.
normal_i2c could be renamed to normal, but I decided not to do so at the
moment, so as to limit the number of changes. This might be done later
as part of the i2c_probe/i2c_detect merge.
Jean Delvare [Tue, 19 Jul 2005 21:57:54 +0000 (23:57 +0200)]
[PATCH] I2C: Separate non-i2c hwmon drivers from i2c-core (6/9)
Kill all isa-related stuff from i2c_detect, it's not used anymore.
This is one major step in the directiom of merging i2c_probe and
i2c_detect. The last obstacle I can think of is the different way forced
addresses work between sensors and non-sensors i2c drivers. I'll deal
with that in a later patchset.
Jean Delvare [Tue, 19 Jul 2005 21:56:35 +0000 (23:56 +0200)]
[PATCH] I2C: Separate non-i2c hwmon drivers from i2c-core (5/9)
Call the ISA chip drivers detection function directly instead of relying
on i2c_detect. The net effect is that address lists won't be handled
anymore, but they were mostly useless in the ISA case anyway (pc87360,
smsc47m1, smsc47b397 had already dropped them).
We don't need to handle multiple devices, all we may need is a way to
force a given address instead of the original one (some drivers already
do: sis5595, via686a, w83627hf), and, for drivers supporting multiple
chips, a way to force one given kind. All this may be added later on
demand, but I actually don't think there will be much demand.
Jean Delvare [Tue, 19 Jul 2005 21:53:07 +0000 (23:53 +0200)]
[PATCH] I2C: Separate non-i2c hwmon drivers from i2c-core (4/9)
All ISA hardware monitoring drivers (including hybrid drivers) now have
a hard dependency on i2c-isa, so they must select I2C_ISA. As a result,
CONFIG_I2C_ISA doesn't need to be left visible to the user. The good
thing here is that users will stop complaining that some driver doesn't
work just because they forgot to compile or load i2c-isa.
At this point, all drivers are working again and the cleanup phase can
begin.
Jean Delvare [Tue, 19 Jul 2005 21:51:07 +0000 (23:51 +0200)]
[PATCH] I2C: Separate non-i2c hwmon drivers from i2c-core (3/9)
Convert the 10 ISA hardware monitoring drivers (it87, lm78, pc87360,
sis5595, smsc47b397, smsc47m1, via686a, w83627hf, w83627ehf, w83781d) to
explicitely register with i2c-isa. For hybrid drivers (it87, lm78,
w83781d), we now have two separate instances of i2c_driver, one for the
I2C interface of the chip, and one for ISA interface. In the long run,
the one for ISA will be replaced with a different driver type.
At this point, all drivers are working again, except for missing
dependencies in Kconfig.
Jean Delvare [Tue, 19 Jul 2005 21:48:43 +0000 (23:48 +0200)]
[PATCH] I2C: Separate non-i2c hwmon drivers from i2c-core (2/9)
Convert i2c-isa from a dumb i2c_adapter into a pseudo i2c-core for ISA
hardware monitoring drivers. The isa i2c_adapter is no more registered
with i2c-core, drivers have to explicitely connect to it using the new
i2c_isa_{add,del}_driver interface.
At this point, all ISA chip drivers are useless, because they still
register with i2c-core in the hope i2c-isa is registered there as well,
but it isn't anymore.
The fake bus will be named i2c-9191 in sysfs. This is the number it
already had internally in various places, so it's not exactly new,
except that now the number is seen in userspace as well. This shouldn't
be a problem until someone really has 9192 I2C busses in a given system
;)
The fake bus will no more show in "i2cdetect -l", as it won't be seen by
i2c-dev anymore (not being registered with i2c-core), which is a good
thing, as i2cdetect/i2cdump/i2cset cannot operate on this fake bus
anyway.
Mark M. Hoffman [Sat, 16 Jul 2005 01:39:18 +0000 (21:39 -0400)]
[PATCH] I2C hwmon: add hwmon sysfs class to drivers
This patch modifies sensors chip drivers to make use of the new
sysfs class "hwmon".
Signed-off-by: Mark M. Hoffman <mhoffman@lightlink.com> Signed-off-by: Jean Delvare <khali@linux-fr.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Mark M. Hoffman [Sat, 16 Jul 2005 01:38:08 +0000 (21:38 -0400)]
[PATCH] I2C hwmon: hwmon sysfs class
This patch adds the sysfs class "hwmon" for use by hardware monitoring
(sensors) chip drivers. It also fixes up the related Kconfig/Makefile
bits.
Signed-off-by: Mark M. Hoffman <mhoffman@lightlink.com> Signed-off-by: Jean Delvare <khali@linux-fr.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Jean Delvare [Wed, 27 Jul 2005 19:30:16 +0000 (21:30 +0200)]
[PATCH] hwmon: kill client name lm78-j
Drop the separate client name for the LM78-J chip. This is really
only a later revision of the LM78, with almost no difference and
no difference the driver handles in any case.
This was the only client name that had a dash in it, and special care
had to be taken in libsensors because of it. As we plan to write a new
library soon, I'd like to get rid of this exception before we do.
As a nice side effect, it saves 876 bytes in lm78.ko.
Signed-off-by: Jean Delvare <khali@linux-fr.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Jean Delvare [Wed, 27 Jul 2005 19:28:28 +0000 (21:28 +0200)]
[PATCH] hwmon: soften lm75 initialization
The LM75 initialization is a bit agressive, it arbitrarily reconfigures
the chip. Make it only change the bit it needs. This is a port from
the 2.4 kernel version of the driver (lm_sensors).
Signed-off-by: Jean Delvare <khali@linux-fr.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
attached is a small patch that removes unused code from i2c-nforce2 and
adds a single debug message. The patch is against 2.6.13-rc3-mm1.
I have tested the patch with 2.6.13-rc3: compiles cleanly and works as
without the patch (as expected).
This is an update to the max6875 driver.
It no longer does any detection, so the address must be forced on module load.
It only makes available the user EEPROM (read-only).
This patch is based off 2.6.13-rc2-mm2.
Signed-off-by: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
On s390 the lock value used for spinlocks consists of the lower 32 bits of the
PSW that holds the lock. If this address happens to be on a four gigabyte
boundary the lock is left unlocked. This allows other cpus to grab the same
lock and enter a lock protected code path concurrently. In theory this can
happen if the vmalloc area for the code of a module crosses a 4 GB boundary.
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
There is a race in pfault_interrupt. That function gets called two times for
each pfault notification. Once with a subcode of 0 to indicate that a real
page is not available and once with a subcode of 0x80 to indicate that the
page is present again.
Since the two external interrupts can be delivered on two different cpus the
order in which the two calls are made is unpredictable. It is possible that
the subcode 0x80 interrupt is completed before the subcode 0x00 interrupt has
done the wake_up() call.
To avoid calling wake_up() on an already removed task structure proper task
structure reference counting is needed. Increase the reference counter in the
subcode 0x00 interrupt before setting pfault_wait to zero and return the
reference after the wake_up call.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
- Collect the irb at the correct subchannel when waiting for the clear
interrupt during subchannel cleaning befor reIPL - don't stop at the first
interrupt that comes in.
- Change "extern __inline__" to "static inline".
- Remove unneeded qdio includes.
Signed-off-by: Cornelia Huck <cohuck@de.ibm.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Add support for diag 250 access to dasd devices for 64 bit kernels. In
addition fix detach/attach for diag disks. The VM control block needs to get
recreated by a call to mdsk_init_io.
Signed-off-by: Horst Hummel <horst.hummel@de.ibm.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Reintroduce a read-only copy of the devmap features in the device struct.
This is necessary to solve a deadlock on the dasd_devmap_lock which is
acquired by dasd_get_features called from the dasd tasklet. The current
implementation of devmap doesn't allow to call any devmap function from
interrupt or softirq context.
Signed-off-by: Horst Hummel <horst.hummel@de.ibm.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
The new machine check handler still has a few bugs.
1) The system entry time has to be stored in the machine check handler,
2) the machine check return psw may not be stored at the usual place
because it might overwrite the return psw of the interrupted context,
3) the return address for the call to s390_handle_mcck in the i/o interrupt
handler is not correct,
4) the system call cleanup has to take the different save area of the
machine check handler into account,
5) the machine check handler may not call UPDATE_VTIME before
CREATE_STACK_FRAME, and
6) the io leave path needs a critical section cleanup to make sure that the
TIF_MCCK_PENDING bit is really checked before switching back to user space.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Bodo Stroesser [Sat, 3 Sep 2005 22:57:51 +0000 (15:57 -0700)]
[PATCH] uml: fix advanced sysemu check
cleanup and fix the check for advanced sysemu (PTRACE_SYSEMU_SINGLESTEP
option)
Signed-off-by: Bodo Stroesser <bstroesser@fujitsu-siemens.com> Signed-off-by: Jeff Dike <jdike@addtoit.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Bodo Stroesser [Sat, 3 Sep 2005 22:57:51 +0000 (15:57 -0700)]
[PATCH] uml: allow host capability usage to be disabled
Add new cmdline setups:
- noprocmm
- noptracefaultinfo
In case of testing, they can be used to switch off usage of
/proc/mm and PTRACE_FAULTINFO independently.
Signed-off-by: Bodo Stroesser <bstroesser@fujitsu-siemens.com> Signed-off-by: Jeff Dike <jdike@addtoit.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Bodo Stroesser [Sat, 3 Sep 2005 22:57:50 +0000 (15:57 -0700)]
[PATCH] uml: skas0 stubs now check system call return values
Change syscall-stub's data to include a "expected retval".
Stub now checks syscalls retval and aborts execution of syscall list, if
retval != expected retval.
run_syscall_stub prints the data of the failed syscall, using the data pointer
and retval written by the stub to the beginning of the stack.
one_syscall_stub is removed, to simplify code, because only some instructions
are saved by one_syscall_stub, no host-syscall.
Using the stub with additional data (modify_ldt via stub)
is prepared also.
Signed-off-by: Bodo Stroesser <bstroesser@fujitsu-siemens.com> Signed-off-by: Jeff Dike <jdike@addtoit.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Bodo Stroesser [Sat, 3 Sep 2005 22:57:49 +0000 (15:57 -0700)]
[PATCH] uml: increase granularity of host capability checking
This change enables SKAS0/SKAS3 to work with all combinations of /proc/mm and
PTRACE_FAULTINFO being available or not.
Also it changes the initialization of proc_mm and ptrace_faultinfo slightly,
to ease forcing SKAS0 on a patched host. Forcing UML to run without /proc/mm
or PTRACE_FAULTINFO by cmdline parameter can be implemented with a setup
resetting the related variable.
Signed-off-by: Bodo Stroesser <bstroesser@fujitsu-siemens.com> Signed-off-by: Jeff Dike <jdike@addtoit.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Jeff Dike [Sat, 3 Sep 2005 22:57:46 +0000 (15:57 -0700)]
[PATCH] uml: add host AIO support to block driver
This adds AIO support to the ubd driver.
The driver breaks a struct request into IO requests to the host, based on the
hardware segments in the request and on any COW blocks covered by the request.
The ubd IO thread is gone, since there is now an equivalent thread in the AIO
module.
There is provision for multiple outstanding requests now. Requests aren't
retired until all pieces of it have been completed. The AIO requests have a
shared count, which is decremented as IO operations come in until it reaches
0. This can be possibly moved to the request struct - haven't looked at this
yet.
Signed-off-by: Jeff Dike <jdike@addtoit.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Jeff Dike [Sat, 3 Sep 2005 22:57:45 +0000 (15:57 -0700)]
[PATCH] uml: use host AIO support
This patch makes UML use host AIO support when it (and
/usr/include/linux/aio_abi.h) are present. This is only the support, with no
consumers - a consumer is coming in the next patch.
Signed-off-by: Jeff Dike <jdike@addtoit.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Al Viro [Sat, 3 Sep 2005 22:57:43 +0000 (15:57 -0700)]
[PATCH] uml: build cleanups
Added missing include list to uml AFLAGS
Killed magic for stubs. [So] - it was needed only because of messed AFLAGS
Switched segv_stubs.c to kernel CFLAGS sans profile, instead of user ones
Killed STUBS_CFLAGS - it's not needed and the only remaining use had been
gratitious - it only polluted CFLAGS
Signed-off-by: Al Viro <viro@parcelfarce.linux.theplanet.co.uk> Signed-off-by: Jeff Dike <jdike@addtoit.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Jeff Dike [Sat, 3 Sep 2005 22:57:41 +0000 (15:57 -0700)]
[PATCH] uml: merge duplicated page table code
There is a lot of code which is duplicated between the 2 and 3 level
implementation, with the only difference that the 3-level implementation is a
bit more generalized (instead of accessing directly pte_t.pte, it uses the
appropriate access macros).
So this code is joined together.
As obvious, a "core code nice cleanup" is not a "stability-friendly patch" so
usual care applies.
Signed-off-by: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it> Signed-off-by: Jeff Dike <jdike@addtoit.com> Cc: Paolo Giarrusso <blaisorblade@yahoo.it> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Jeff Dike [Sat, 3 Sep 2005 22:57:36 +0000 (15:57 -0700)]
[PATCH] uml: TLB operation batching
This adds VM op batching to skas0. Rather than having a context switch to and
from the userspace stub for each address space change, we write a number of
operations to the stub data page and invoke a different stub which loops over
them and executes them all in one go.
The operations are stored as [ system call number, arg1, arg2, ... ] tuples.
The set is terminated by a system call number of 0. Single operations, i.e.
page faults, are handled in the old way, since that is slightly more
efficient.
For a kernel build, a minority (~1/4) of the operations are part of a set.
These sets averaged ~100 in length, so for this quarter, the context switching
overhead is greatly reduced.
Signed-off-by: Jeff Dike <jdike@addtoit.com> Cc: Paolo Giarrusso <blaisorblade@yahoo.it> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Jeff Dike [Sat, 3 Sep 2005 22:57:34 +0000 (15:57 -0700)]
[PATCH] uml: remove duplicated exports
Al Viro <viro@parcelfarce.linux.theplanet.co.uk> spotted a bunch of duplicated
exports - this removes them.
Signed-off-by: Jeff Dike <jdike@addtoit.com> Cc: Paolo Giarrusso <blaisorblade@yahoo.it> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Jeff Dike [Sat, 3 Sep 2005 22:57:33 +0000 (15:57 -0700)]
[PATCH] uml: mark SMP on UML/x86_64 as broken
Noticed by Al Viro <viro@parcelfarce.linux.theplanet.co.uk> - SMP on x86_64 is
fundamentally broken due to UML's reuse of the host arch's percpu stuff. This
is OK on x86, but the x86_64 pda stuff just won't work for UML.
Signed-off-by: Jeff Dike <jdike@addtoit.com> Cc: Paolo Giarrusso <blaisorblade@yahoo.it> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Al Viro [Sat, 3 Sep 2005 22:57:31 +0000 (15:57 -0700)]
[PATCH] uml: remove libc reference in build
Remove an unneeded reference to libc.
Signed-off-by: Jeff Dike <jdike@addtoit.com> Cc: Paolo Giarrusso <blaisorblade@yahoo.it> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Al Viro [Sat, 3 Sep 2005 22:57:30 +0000 (15:57 -0700)]
[PATCH] uml: build cleanup
Build cleanups
Signed-off-by: Jeff Dike <jdike@addtoit.com> Cc: Paolo Giarrusso <blaisorblade@yahoo.it> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Jeff Dike [Sat, 3 Sep 2005 22:57:29 +0000 (15:57 -0700)]
[PATCH] uml: error path cleanup
This cleans up the error path in ubd_open, causing it now to call ubd_close
appropriately when something fails.
Signed-off-by: Jeff Dike <jdike@addtoit.com> Cc: Paolo Giarrusso <blaisorblade@yahoo.it> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Avoid chomping low bits of address for functions doing it by themselves,
fix whitespace, add a correctness checking.
I did this for remap-file-pages protection support, it was useful on its
own too.
Signed-off-by: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it> Cc: Jeff Dike <jdike@addtoit.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
[PATCH] uml: fixes performance regression in activate_mm and thus exec()
Normally, activate_mm() is called from exec(), and thus it used to be a
no-op because we use a completely new "MM context" on the host (for
instance, a new process), and so we didn't need to flush any "TLB entries"
(which for us are the set of memory mappings for the host process from the
virtual "RAM" file).
Kernel threads, instead, are usually handled in a different way. So, when
for AIO we call use_mm(), things used to break and so Benjamin implemented
activate_mm(). However, that is only needed for AIO, and could slow down
exec() inside UML, so be smart: detect being called for AIO (via
PF_BORROWED_MM) and do the full flush only in that situation.
Comment also the caller so that people won't go breaking UML without
noticing. I also rely on the caller's locks for testing current->flags.
Signed-off-by: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it> CC: Benjamin LaHaise <bcrl@kvack.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Bodo Stroesser [Sat, 3 Sep 2005 22:57:24 +0000 (15:57 -0700)]
[PATCH] uml: fix SIGWINCH handler race while waiting for signals.
If a SIGWINCH comes in, while winch_thread() isn't waiting in wait(),
winch_thread could miss signals. It isn't very probable, that anyone will
see this causing trouble, as it would need a very special timing, that a
missed SIGWINCH results in a wrong window size.
So, this is a minor problem. But why not fix, as it can be done so easy?
Signed-off-by: Bodo Stroesser <bstroesser@fujitsu-siemens.com> Signed-off-by: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it> Cc: Jeff Dike <jdike@addtoit.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Apparently, GDB gets confused when we do an execvp() on ourselves.
Since it's simply done to allocate further space for command line arguments
(which we'll use to allow gathering the startup command line for guest
processes through the host), allow the user to disable that to get a
debuggable UML binary.
Signed-off-by: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it> Cc: Jeff Dike <jdike@addtoit.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
As a follow-up to "UML Support - Ptrace: adds the host SYSEMU support, for
UML and general usage" (i.e. uml-support-* in current mm).
Avoid unconditionally jumping to work_pending and code copying, just reuse
the already existing resume_userspace path.
One interesting note, from Charles P. Wright, suggested that the API is
improvable with no downsides for UML (except that it will have to support
yet another host API, since dropping support for the current API, for UML,
is not reasonable from users' point of view).
Signed-off-by: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it> CC: Charles P. Wright <cwright@cs.sunysb.edu> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it>
This is simply an adjustment for "Ptrace - i386: fix Syscall Audit interaction
with singlestep" to work on top of SYSEMU patches, too. On this patch, I have
some doubts: I wonder why we need to alter that way ptrace_disable().
I left the patch this way because it has been extensively tested, but I don't
understand the reason.
The current PTRACE_DETACH handling simply clears child->ptrace; actually this
is not enough because entry.S just looks at the thread_flags; actually,
do_syscall_trace checks current->ptrace but I don't think depending on that is
good, at least for performance, so I think the clearing is done elsewhere.
For instance, on PTRACE_CONT it's done, but doing PTRACE_DETACH without
PTRACE_CONT is possible (and happens when gdb crashes and one kills it
manually).
Signed-off-by: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it> CC: Roland McGrath <roland@redhat.com> Cc: Jeff Dike <jdike@addtoit.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Bodo Stroesser [Sat, 3 Sep 2005 22:57:20 +0000 (15:57 -0700)]
[PATCH] Uml support: add PTRACE_SYSEMU_SINGLESTEP option to i386
This patch implements the new ptrace option PTRACE_SYSEMU_SINGLESTEP, which
can be used by UML to singlestep a process: it will receive SINGLESTEP
interceptions for normal instructions and syscalls, but syscall execution will
be skipped just like with PTRACE_SYSEMU.
Signed-off-by: Bodo Stroesser <bstroesser@fujitsu-siemens.com> Signed-off-by: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it> Cc: Jeff Dike <jdike@addtoit.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Bodo Stroesser [Sat, 3 Sep 2005 22:57:19 +0000 (15:57 -0700)]
[PATCH] Uml support: reorganize PTRACE_SYSEMU support
With this patch, we change the way we handle switching from PTRACE_SYSEMU to
PTRACE_{SINGLESTEP,SYSCALL}, to free TIF_SYSCALL_EMU from double use as a
preparation for PTRACE_SYSEMU_SINGLESTEP extension, without changing the
behavior of the host kernel.
Signed-off-by: Bodo Stroesser <bstroesser@fujitsu-siemens.com> Signed-off-by: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it> Cc: Jeff Dike <jdike@addtoit.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
[PATCH] UML Support - Ptrace: adds the host SYSEMU support, for UML and general usage
Jeff Dike <jdike@addtoit.com>,
Paolo 'Blaisorblade' Giarrusso <blaisorblade_spam@yahoo.it>,
Bodo Stroesser <bstroesser@fujitsu-siemens.com>
Adds a new ptrace(2) mode, called PTRACE_SYSEMU, resembling PTRACE_SYSCALL
except that the kernel does not execute the requested syscall; this is useful
to improve performance for virtual environments, like UML, which want to run
the syscall on their own.
In fact, using PTRACE_SYSCALL means stopping child execution twice, on entry
and on exit, and each time you also have two context switches; with SYSEMU you
avoid the 2nd stop and so save two context switches per syscall.
Also, some architectures don't have support in the host for changing the
syscall number via ptrace(), which is currently needed to skip syscall
execution (UML turns any syscall into getpid() to avoid it being executed on
the host). Fixing that is hard, while SYSEMU is easier to implement.
* This version of the patch includes some suggestions of Jeff Dike to avoid
adding any instructions to the syscall fast path, plus some other little
changes, by myself, to make it work even when the syscall is executed with
SYSENTER (but I'm unsure about them). It has been widely tested for quite a
lot of time.
* Various fixed were included to handle the various switches between
various states, i.e. when for instance a syscall entry is traced with one of
PT_SYSCALL / _SYSEMU / _SINGLESTEP and another one is used on exit.
Basically, this is done by remembering which one of them was used even after
the call to ptrace_notify().
* We're combining TIF_SYSCALL_EMU with TIF_SYSCALL_TRACE or TIF_SINGLESTEP
to make do_syscall_trace() notice that the current syscall was started with
SYSEMU on entry, so that no notification ought to be done in the exit path;
this is a bit of a hack, so this problem is solved in another way in next
patches.
* Also, the effects of the patch:
"Ptrace - i386: fix Syscall Audit interaction with singlestep"
are cancelled; they are restored back in the last patch of this series.
Detailed descriptions of the patches doing this kind of processing follow (but
I've already summed everything up).
* Fix behaviour when changing interception kind #1.
In do_syscall_trace(), we check the status of the TIF_SYSCALL_EMU flag
only after doing the debugger notification; but the debugger might have
changed the status of this flag because he continued execution with
PTRACE_SYSCALL, so this is wrong. This patch fixes it by saving the flag
status before calling ptrace_notify().
* Fix behaviour when changing interception kind #2:
avoid intercepting syscall on return when using SYSCALL again.
A guest process switching from using PTRACE_SYSEMU to PTRACE_SYSCALL
crashes.
The problem is in arch/i386/kernel/entry.S. The current SYSEMU patch
inhibits the syscall-handler to be called, but does not prevent
do_syscall_trace() to be called after this for syscall completion
interception.
The appended patch fixes this. It reuses the flag TIF_SYSCALL_EMU to
remember "we come from PTRACE_SYSEMU and now are in PTRACE_SYSCALL", since
the flag is unused in the depicted situation.
* Fix behaviour when changing interception kind #3:
avoid intercepting syscall on return when using SINGLESTEP.
When testing 2.6.9 and the skas3.v6 patch, with my latest patch and had
problems with singlestepping on UML in SKAS with SYSEMU. It looped
receiving SIGTRAPs without moving forward. EIP of the traced process was
the same for all SIGTRAPs.
What's missing is to handle switching from PTRACE_SYSCALL_EMU to
PTRACE_SINGLESTEP in a way very similar to what is done for the change from
PTRACE_SYSCALL_EMU to PTRACE_SYSCALL_TRACE.
I.e., after calling ptrace(PTRACE_SYSEMU), on the return path, the debugger is
notified and then wake ups the process; the syscall is executed (or skipped,
when do_syscall_trace() returns 0, i.e. when using PTRACE_SYSEMU), and
do_syscall_trace() is called again. Since we are on the return path of a
SYSEMU'd syscall, if the wake up is performed through ptrace(PTRACE_SYSCALL),
we must still avoid notifying the parent of the syscall exit. Now, this
behaviour is extended even to resuming with PTRACE_SINGLESTEP.
Signed-off-by: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it> Cc: Jeff Dike <jdike@addtoit.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Bodo Stroesser [Sat, 3 Sep 2005 22:57:13 +0000 (15:57 -0700)]
[PATCH] Ptrace/i386: fix "syscall audit" interaction with singlestep
Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it>
Avoid giving two traps for singlestep instead of one, when syscall auditing is
enabled.
In fact no singlestep trap is sent on syscall entry, only on syscall exit, as
can be seen in entry.S:
# Note that in this mask _TIF_SINGLESTEP is not tested !!! <<<<<<<<<<<<<<
testb $(_TIF_SYSCALL_TRACE|_TIF_SYSCALL_AUDIT|_TIF_SECCOMP),TI_flags(%ebp)
jnz syscall_trace_entry
...
syscall_trace_entry:
...
call do_syscall_trace
But auditing a SINGLESTEP'ed process causes do_syscall_trace to be called, so
the tracer will get one more trap on the syscall entry path, which it
shouldn't.
Signed-off-by: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it> CC: Roland McGrath <roland@redhat.com> Cc: Jeff Dike <jdike@addtoit.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Jeff Dike [Sat, 3 Sep 2005 22:57:12 +0000 (15:57 -0700)]
[PATCH] uml: Rename Kconfig files to be like the other arches
To the extent that sub-Kconfig files exist elsewhere in the tree, they are
named Kconfig.foo, rather than the Kconfig_foo that UML has. This patch
brings the names in line with the rest of the tree.
Signed-off-by: Jeff Dike <jdike@addtoit.com> Cc: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
projects / linux-beck.git / log