A kernel panic may occur during sending or receiving network packets
on a machine without adapter interrupts since commit d36deae.
The bug is triggered by writing to the shared indicator address which
is set to 0 if the machine doesn't have adapter interrupts.
Make the reading and setting of the shared indicator dependent on the
adapter interrupt feature and while at it move the code to the
file containing the adapter interrupt related code.
Thanks to Jan Jaeger for tracking this down.
Reported-by: Jan Jaeger <jan.jaeger@westnet.com.au>
Tested-by: Jan Jaeger <jan.jaeger@westnet.com.au>
Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
The multicast poll check for the outbound queue is redundant since
3d6c76f "[S390] qdio: outbound tasklet scan threshold". Remove the check.
Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Add a timestamp per queue and update the timestamp when the queue is
scanned. Add the queue timestamps and the timestamp of the last
adapter interrupt to the debugfs output. The timestamps are useful
for debugging stall conditions.
Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
This patch ensures that signal adapter commands are issued if they are
indicated to be required.
Signed-off-by: Einar Lueck <elelueck@de.ibm.com>
Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Frank Blaschka <frank.blaschka@de.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch introduces support for asynchronous delivery of storage blocks for
Hipersockets. Upper layers may exploit this functionality to reuse SBALs for
which the delivery status is still pending.
Signed-off-by: Einar Lueck <elelueck@de.ibm.com>
Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Frank Blaschka <frank.blaschka@de.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The SIGA-W may return with the busy bit set which means the device was
blocked. The busy loop which retries the SIGA-W for 100us may not be
long enough when running under a heavily loaded hypervisor.
Extend the retry mechanism by adding a longer second stage which retries
the SIGA-W for up to 10s. In difference to the first retry loop the second
stage is using mdelay to stop the cpu between the retries and thereby
avoid additional preassure in on the hypervisor.
If the second stage retry is successfull a device reset is avoided.
Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Simplify the SIGA sync code and add unlikely annotations. In polling mode
SBALs may be accessed without interrupt, so call SIGA sync before every scan.
Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
HiperSocket devices only use one SBAL per qdio call without the enhanced SIGA
feature. Since that feature is currently not used remove it from the qdio code
so the compiler can generate better code for the HiperSocket outbound path.
While at it mark the SIGA error conditions as unlikely.
Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
If QIOASSIST is enabled for a qdio device the SIGA instruction requires
a modified function code. This function code modifier was missing for
SIGA-R and SIGA-S which can lead to a kernel panic caused by an
operand exception.
Cc: stable@kernel.org
Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Add a counter for outbound queue full events to the qdio statistics.
Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Introduce a scan treshold for the qdio outbound queues. By setting the
threshold the driver can tell qdio after how much used SBALs qdio
should schedule the outbound tasklet that scans the queue for finished
SBALs. The threshold is specific by the drivers because a
Hipersockets device is much faster in utilizing outbound buffers than a
ZFCP or OSA device.
The default values after how many used SBALs the tasklet should run are:
OSA: > 31 SBALs
Hipersockets: > 7 SBALs
zfcp: > 55 SBALs
Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
If the shared indicator is used the following race leads to
an inbound stall:
Device CPU0 CPU1
========================================================
non-shared DSCI =>1
ALSI => 1
Thin INT
ALSI => 0
non-shared DSCI
tasklets scheduled
shared DSCI => 1
ALSI => 1
shared DSCI => 0
ALSI ? -> set
Thin INT
ALSI => 0
ALSI was set,
shared DSCI => 1
After that no more interrupts occur because the DSCI is still set.
Fix that race by only resetting the shared DSCI if it was actually
set so the tasklets for all shared devices are scheduled and will
run after the interrupt.
Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Extend the qdio API to allow polling in the upper-layer driver. This
is needed by qeth to use NAPI.
To use the new interface the upper-layer driver must specify the
queue_start_poll(). This callback is used to signal the upper-layer
driver that is has initiative and must process the inbound queue by
calling qdio_get_next_buffers(). If the upper-layer driver wants to
stop polling it calls qdio_start_irq().
Since adapter interrupts are not completely stoppable qdio implements
a software bit QDIO_QUEUE_IRQS_DISABLED to safely disable interrupts for an
input queue.
The old interface is preserved and will be used as is by zfcp.
Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Frank Blaschka <frank.blaschka@de.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Don't convert timestamps to microseconds, use timestamps returned by
get_clock() directly.
Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Remove memset hack that relied on the layout of struct qdio_q
to avoid deletion of the slib pointer.
Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Remove a memset hack that relied on the internal layout of the
qdio_irq struct and move the per device statistics data into an own
cache line to avoid cache line bashing between the inbound and the
outbound queue tasklets. Also reduce the number of allocated queues
from 32 to 4 which is the current maximum. That saves a cache line
in struct qdio_irq.
Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Add counters for the number of processed SBALs. The numbers summarize
how many SBALs were processed at each queue scan and indicate the
utilization of the queue. Furthermore the number of unsuccessfull
queue scans, SBAL errors and the total number of processed
SBALs are accounted.
Also regroup struct qdio_q to move read-mostly and write-mostly data
into different cachelines.
Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Revamp the qdio performance statistics and move them from procfs to
debugfs using the seq_file interface. Since the statistics are not
intended for the general user the removal of /proc/qdio_perf should
not surprise anyone.
The per device statistics are disabled by default, writing 1 to
/<debugfs mountpoint>/qdio/<device bus ID>/statistics enables the
statistics for the given device.
Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
The number of qdio debugfs entries was limited. Remove this limit
and group the queue files in a per device directory.
Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Move the adapter interrupt tasklet function to the qdio main code
since all the functions used by the tasklet are located there.
Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
The inbound and outbound handlers are nearly identical if the outbound
handler uses first_to_check as end index instead of last_move. Since both
values are identical at that point the handlers can be merged.
Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Errors from SIGA instructions are stored in the per queue qdio_error
and reported back when the queue handler is called. That opens a race
when multiple error conditions occur simultanously.
Report SIGA errors immediately in the return value of do_QDIO so the
upper layer can react and SIGA errors no longer interfere with other
errors.
Move the SIGA error handling in qeth from the outbound handler to
qeth_flush_buffers.
Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
The index value that indicated that the input queue moved was also used to
store the index of the first acknowledged buffer. For non-qebsm only the
newest buffer is acknowledged which may be different from the last move index
so two seperate values are needed to track the input queue.
Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Hipersocket connections can encounter temporary busy conditions.
In case of the busy bit set we retry the SIGA operation immediatelly.
If the busy condition still persists after 100 ms we fail and report
the error to the upper layer. The second stage retry logic is removed.
In case of ongoing busy conditions the upper layer needs to reset the
connection.
The reporting of a SIGA error is now done synchronously to allow the
network driver to requeue the buffers. Also no error trace is created
for the temporary SIGA errors so the error message view is not flooded.
Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
- Use automatic acknowledgement of incoming buffers in QEBSM mode
- Move ACK for non-QEBSM mode always to the newest buffer to prevent
a race with qdio_stop_polling
- Remove the polling spinlock, the upper layer drivers return new buffers
in the same code path and could not run in parallel
- Don't flood the error log in case of no-target-buffer-empty
- In handle_inbound we check if we would overwrite an ACK'ed buffer, if so
advance the pointer to the oldest ACK'ed buffer so we don't overwrite an
empty buffer in qdio_stop_polling
Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
- make qdio_trace a per device view
- remove s390dbf exceptions
- remove CONFIG_QDIO_DEBUG, not needed anymore if we check for the level
before calling sprintf
- use snprintf for dbf entries
- add start markers to see if the dbf view wrapped
- add a global error view for all queues
Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
qeth needs to get the port count information before
qdio has allocated a page for the chsc operation.
Extend qdio_get_ssqd_desc() to store the data in the
specified structure.
Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Add support for z10 HiperSockets multiwrite SBALs on output
queues. This is used on LPAR with EDDP enabled devices.
Signed-off-by: Klaus-Dieter Wacker <kdwacker@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
If an asynchronous HiperSockets queue runs full, no further packet
can be sent. In this case the next initiative to give transmitted
skbs back to the stack is triggered only by a 10-seconds qdio timer.
This timer has been introduced for low multicast traffic scenarios
to guarantee freeing of skbs in a limited amount of time. For high
HiperSocket multicast traffic scenarios progress checking on the
outbound queue should be enforced by tasklet rescheduling.
Signed-off-by: Ursula Braun <ursula.braun@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
This patch adds a driver for subchannels of type chsc.
A device /dev/chsc is created which may be used to issue ioctls to:
- obtain information about the machine's I/O configuration
- dynamically change the machine's I/O configuration via
asynchronous chsc commands
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Replace the numeric values for I/O interruption subclass usage
with abstract definitions and collect them all in asm/isc.h.
This gives us a better overview of which iscs are actually used
and makes it possible to better spread out isc usage in the
future.
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Next generation of OSA adapters allows retrieval of further self-describing
infos. This is the preparational infrastructure patch for further exploitation
in the qeth driver.
Signed-off-by: Ursula Braun <braunu@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Current code in qdio_activate waits for at least 5 seconds
until it returns. It may return earlier if an error occurs,
but not if everything is ok. This large timeout value
became visible with commit dfa77f611f
"qdio: set QDIO_ACTIVATE_TIMEOUT to 5s", which intended to
fix the timeout value which was zero. In turn setting an
FCP adapter online took 5 seconds.
In practice waiting for 5ms before continuing is sufficient
as pointed out by Utz Bacher and Cornelia Huck.
Cc: Utz Bacher <utz.bacher@de.ibm.com>
Cc: Jan Glauber <jan.glauber@de.ibm.com>
Cc: Ursula Braun <braunu@de.ibm.com>
Cc: Martin Peschke <mp3@de.ibm.com>
Acked-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Current definition of QDIO_ACTIVATE_TIMEOUT results in value 0.
Thus it may cause endless wait in function qdio_activate().
Signed-off-by: Ursula Braun <braunu@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Free sent skbs in some finite amount of time. Affected are
asynchronous queue of Hipersockets devices and the output
queues of all eth-devices respectively.
Signed-off-by: Ursula Braun <braunu@de.ibm.com>
Signed-off-by: Frank Pavlic <fpavlic@de.ibm.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
Use atomic_t/atomic64_t to make qdio performance statistics smp safe.
Remove temporarily calculation of "total time of inbound actions".
Signed-off-by: Ursula Braun <braunu@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Remove CONFIG_QETH_PERF_STATS and use a sysfs attribute instead.
We want to have the ability to turn the statistics on/off at runtime.
Signed-off-by: Ursula Braun <braunu@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Add some prefixes to constands defined in drivers/s390/net/qdio.h
and drivers/s390/lcs.h to make it possible to include the three
header files drivers/s390/net/qeth.h, drivers/s390/net/qdio.h and
drivers/net/s390/lcs.h in one C file. This is required for the
patch that generates the kerntypes.o file for use by lcrash.
Signed-off-by: Ursula Braun <braunu@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Major cleanup of all s390 inline assemblies. They now have a common
coding style. Quite a few have been shortened, mainly by using register
asm variables. Use of the EX_TABLE macro helps as well. The atomic ops,
bit ops and locking inlines new use the Q-constraint if a newer gcc
is used. That results in slightly better code.
Thanks to Christian Borntraeger for proof reading the changes.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Use do { } while (0) constructs instead of empty defines to avoid
subtle compile bugs.
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
- Remove all CVS generated information like e.g. revision IDs from
drivers/s390 and include/asm-s390 (none present in arch/s390).
- Add newline at end of arch/s390/lib/Makefile to avoid diff message.
Acked-by: Andreas Herrmann <aherrman@de.ibm.com>
Acked-by: Frank Pavlic <pavlic@de.ibm.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Sanitize some s390 Kconfig options. We have ARCH_S390, ARCH_S390X,
ARCH_S390_31, 64BIT, S390_SUPPORT and COMPAT. Replace these 6 options by
S390, 64BIT and COMPAT.
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>
This patch introduces a struct subchannel_id containing the subchannel number
(formerly referred to as "irq") and switches code formerly relying on the
subchannel number over to it.
While we're touching inline assemblies anyway, make sure they have correct
memory constraints.
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>
New feature V=V qdio pass-through.
QDIO and HiperSockets processing in z/VM V=V guest environments (as well as
V=R with z/VM running in LPAR mode) requires shadowing of all QDIO
architecture queue elements. Especially the shadowing of SBALs and SLSBs
structures in the hypervisor, and the need to issue SIGA SYNC operations to
observe state changes, eventually causes significant CPU processing overhead
in the hypervisor.
The QDIO pass-through support for V=V guests avoids the shadowing of SBALs and
SLSBs. This significantly reduces the hypervisor overhead for QDIO based I/O.
Signed-off-by: Frank Pavlic <pavlic@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>
This patch changes the memory allocation method for the s390 debug feature.
Trace buffers had been allocated using the get_free_pages() function before.
Therefore it was not possible to get big memory areas in a running system due
to memory fragmentation. Now the trace buffers are subdivided into several
subbuffers with pagesize. Therefore it is now possible to allocate more
memory for the trace buffers and more trace records can be written.
In addition to that, dynamic specification of the size of the trace buffers is
implemented. It is now possible to change the size of a trace buffer using a
new debugfs file instance. When writing a number into this file, the trace
buffer size is changed to 'number * pagesize'.
In the past all the traces could be obtained from userspace by accessing files
in the "proc" filesystem. Now with debugfs we have a new filesystem which
should be used for debugging purposes. This patch moves the debug feature
from procfs to debugfs.
Since the interface of debug_register() changed, all device drivers, which use
the debug feature had to be adjusted.
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>
Provide an easy way to define a non-zero storage key at compile time. This is
useful for debugging purposes.
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>
Initial git repository build. I'm not bothering with the full history,
even though we have it. We can create a separate "historical" git
archive of that later if we want to, and in the meantime it's about
3.2GB when imported into git - space that would just make the early
git days unnecessarily complicated, when we don't have a lot of good
infrastructure for it.
Let it rip!
