Note: This is compile only tested as I have no access to the hw.
This variable was not used in qib_sdma.c but in qib_iba7322.c. Declaring it
there, as static, saves 56 bytes.
add/remove: 0/2 grow/shrink: 0/0 up/down: 0/-144 (-144)
Function old new delta
qib_sdma_state_names 56 - -56
qib_sdma_event_names 88 - -88
Total: Before=2874565, After=2874421, chg -0.01%
Signed-off-by: Hernán Gonzalez <hernan@vanguardiasur.com.ar>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Note: This is compile only tested as I have no access to the hw.
This variable was not used anywhere in the code. Removing it saves 88
bytes.
add/remove: 0/1 grow/shrink: 0/0 up/down: 0/-88 (-88)
Function old new delta
qib_sdma_event_names 88 - -88
Total: Before=2874565, After=2874477, chg -0.00%
Signed-off-by: Hernán Gonzalez <hernan@vanguardiasur.com.ar>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
The get_unit_name() function crafts a string based on the device name
and the device unit number. It then stores this in a static variable.
This has concurrency issues as can be seen with this log:
hfi1 0000:02:00.0: hfi1_1: read_idle_message: read idle message 0x203
hfi1 0000:01:00.0: hfi1_1: read_idle_message: read idle message 0x203
The PCI device ID (0000:02:00.0 vs. 0000:01:00.0) is correct for the
message, but the device string hfi1_1 is incorrect (it should be
hfi1_0 for the second log message).
Remove get_unit_name() function.
Instead, use the rvt accessor rvt_get_ibdev_name() to get the IB name
string.
Clean up any hfi1_early_xx calls that can now use the new path.
QIB has the same (qib_get_unit_name()) issue. Updating as necessary.
Remove qib_get_unit_name() function.
Update log message that has redundant device name.
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Michael J. Ruhl <michael.j.ruhl@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
rdmavt has a down call to client drivers to retrieve a crafted card
name.
This name should be the IB defined name.
Rather than craft the name each time it is needed, simply retrieve
the IB allocated name from the IB device.
Update the function name to reflect its application.
Clean up driver code to match this change.
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Michael J. Ruhl <michael.j.ruhl@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Conflicts:
drivers/infiniband/hw/cxgb4/cm.c
drivers/infiniband/hw/qib/qib_driver.c
drivers/infiniband/hw/qib/qib_mad.c
There were minor fixups needed in these files. Just minor context diffs
due to patches from independent sources touching the same basic area.
Signed-off-by: Doug Ledford <dledford@redhat.com>
Due to removal of the EEPROM writing code, the qib_inc_eeprom_err()
macro became a no-op. Remove the code that calls it. Since that
change removes all code that reads the eep_st_masks array, also
remove the code that updates that array and the array itself.
References: commit 18c0b82a3e ("IB/qib: Do not write EEPROM")
Signed-off-by: Bart Van Assche <bart.vanassche@wdc.com>
Cc: Mike Marciniszyn <infinipath@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
In preparation for unconditionally passing the struct timer_list pointer to
all timer callbacks, switch to using the new timer_setup() and from_timer()
helper to pass the timer pointer explicitly.
Cc: Mike Marciniszyn <infinipath@intel.com>
Cc: Doug Ledford <dledford@redhat.com>
Cc: Sean Hefty <sean.hefty@intel.com>
Cc: Hal Rosenstock <hal.rosenstock@gmail.com>
Cc: linux-rdma@vger.kernel.org
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Doug Ledford <dledford@redhat.com>
The QIB PCI IRQ code uses an obsolete PCI API. Updating the code to use
the new PCI IRQ API and any necessary changes because of the new API.
Reviewed-by: Sebastian Sanchez <sebastian.sanchez@intel.com>
Signed-off-by: Michael J. Ruhl <michael.j.ruhl@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
pci_enable_msix_range() and pci_disable_msix() have been deprecated.
Updating to the new pci_alloc_irq_vectors() interface.
Reviewed-by: Sebastian Sanchez <sebastian.sanchez@intel.com>
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Michael J. Ruhl <michael.j.ruhl@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
There is no need to have a global qpt_mask as that does not support the
multiple chip model which qib has. Instead rely on the value which
exists already in the device data (dd).
Fixes: 898fa52b4a "IB/qib: Remove qpn, qp tables and related variables from qib"
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Use common header file structs, defines, and accessors
in the drivers. The old declarations are removed.
The repositioning of the includes allows for the removal
of hfi1_message_header and replaces its use with ib_header.
Also corrected are two issues with set_armed_to_active():
- The "packet" parameter is now a pointer as it should have been
- The etype is validated to insure that the header is correct
Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Reviewed-by: Don Hiatt <don.hiatt@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
This would avoid conflict with the functions in hfi1 that have similar
names when both qib and hfi1 drivers are configured to be built into
the kernel. This issue came up in the 0-day build report.
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Harish Chegondi <harish.chegondi@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Modify queue pair functionality in rdmavt will be used instead.
Remove ancillary functions which are being used by modify QP code.
Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Harish Chegondi <harish.chegondi@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Add calls to rcu_read_lock()/rcu_read_unlock() as rvt_lookup_qpn callers
must hold the rcu_read_lock before calling and keep the lock until the
returned qp is no longer in use.
Remove lookaside qp and some qp refcount atomics in the sdma send code
that is redundant with the s_dma_busy refcount, which will also stall
the state processing to the reset state.
Change the qpn hash function to hash_32 which is hash function used
in rvt_lookup_qpn. qpn_hash function would be eliminated in later patches.
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Harish Chegondi <harish.chegondi@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Use the completion queue functionality provided by rdmavt.
Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Harish Chegondi <harish.chegondi@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Rely on rdmavt functions for creation of qp and qp table. Function to
allocate a qpn is still being provided by qib as the algorithm to allocate
a qpn in qib is different from that of the algorithm in rdmavt.
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Harish Chegondi <harish.chegondi@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Remove qp and mr support from qib and use rdmavt. These two changes
cannot be reasonably be split apart into separate patches because they
depend on each other in mulitple places. This paves the way to remove
even more functions in subsequent patches.
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Implement get_card_name and get_pci_dev helper functions for rdmavt
for qib.
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
This patch begins to make use of rdmavt by registering with it and
providing access to the header files. This is just the beginning of
rdmavt support in qib.
Most functionality is still being done in the driver, set flags so that
rdmavt will let qib continue to handle mr, qp, and cq init.
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Commit d4988623cc ("IB/qib: use arch_phys_wc_add()")
adjusted mtrr inititialization to use the new interface.
Unfortunately, the new interface returns a signed
value and the patch tested the unsigned wc_cookie.
Fix the issue by changing the type of wc_cookie to int. For
the success case the ret left at zero to avoid
a warning from the caller. For failure wc_cookie
is used as the ret.
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
This driver already makes use of ioremap_wc() on PIO buffers,
so convert it to use arch_phys_wc_add().
The qib driver uses a mmap() special case for when PAT is
not used, this behaviour used to be determined with a
module parameter but since we have been asked to just
remove that module parameter this checks for the WC cookie,
if not set we can assume PAT was used. If its set we do
what we used to do for the mmap for when MTRR was enabled.
The removal of the module parameter is OK given that Andy
notes that even if users of module parameter are still around
it will not prevent loading of the module on recent kernels.
Cc: Doug Ledford <dledford@redhat.com>
Cc: Toshi Kani <toshi.kani@hp.com>
Cc: Rickard Strandqvist <rickard_strandqvist@spectrumdigital.se>
Cc: Mike Marciniszyn <mike.marciniszyn@intel.com>
Cc: Roland Dreier <roland@purestorage.com>
Cc: Sean Hefty <sean.hefty@intel.com>
Cc: Hal Rosenstock <hal.rosenstock@gmail.com>
Cc: Dennis Dalessandro <dennis.dalessandro@intel.com>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Suresh Siddha <sbsiddha@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Juergen Gross <jgross@suse.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Dave Airlie <airlied@redhat.com>
Cc: Bjorn Helgaas <bhelgaas@google.com>
Cc: Antonino Daplas <adaplas@gmail.com>
Cc: Jean-Christophe Plagniol-Villard <plagnioj@jcrosoft.com>
Cc: Tomi Valkeinen <tomi.valkeinen@ti.com>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Michael S. Tsirkin <mst@redhat.com>
Cc: Stefan Bader <stefan.bader@canonical.com>
Cc: konrad.wilk@oracle.com
Cc: ville.syrjala@linux.intel.com
Cc: david.vrabel@citrix.com
Cc: jbeulich@suse.com
Cc: Roger Pau Monné <roger.pau@citrix.com>
Cc: infinipath@intel.com
Cc: linux-rdma@vger.kernel.org
Cc: linux-fbdev@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Cc: xen-devel@lists.xensource.com
Signed-off-by: Luis R. Rodriguez <mcgrof@suse.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
This changeset removes all the code that allows the driver to write to
the EEPROM and update the recorded error counters and power on hours.
These two stats are unused and writing them exposes a timing risk
which could leave the EEPROM in a bad state preventing further normal
operation of the HCA.
Cc: <stable@vger.kernel.org>
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Mitko Haralanov <mitko.haralanov@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
The counters, unicast_xmit, unicast_rcv, multicast_xmit, multicast_rcv
are now maintained as percpu variables.
The mad code is modified to add a z_ latch so that the percpu counters
monotonically increase with appropriate adjustments in the reset,
read logic to maintain the z_ latch.
This patch also corrects the fact the unitcast_xmit wasn't handled
at all for UC and RC QPs.
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
This patch replaces the dd->int_counter with a percpu counter.
The maintanance of qib_stats.sps_ints and int_counter are
combined into the new counter.
There are two new functions added to read the counter:
- qib_int_counter (for a particular qib_devdata)
- qib_sps_ints (for all HCAs)
A z_int_counter is added to allow the interrupt detection logic
to determine if interrupts have occured since z_int_counter
was "reset".
Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
struct pci_driver qib_driver is only used in qib_init.c. Remove it
from qib.h and make it static in qib_init.c.
Signed-off-by: Paul Bolle <pebolle@tiscali.nl>
Signed-off-by: Roland Dreier <roland@purestorage.com>
1. The code accepts chunks of messages, and splits the chunk into
packets when converting packets into sdma queue entries. Adjacent
packets will use user buffer pages smartly to avoid pinning the
same page multiple times.
2. Instead of discarding all the work when SDMA queue is full, the
work is saved in a pending queue. Whenever there are enough SDMA
queue free entries, pending queue is directly put onto SDMA queue.
3. An interrupt handler is used to progress this pending queue.
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: CQ Tang <cq.tang@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
[ Fixed up sparse warnings. - Roland ]
Signed-off-by: Roland Dreier <roland@purestorage.com>
This patch adds code to log SDMA errors for supportability purposes.
Signed-off-by: Dean Luick <dean.luick@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
This fix changes the opcode relative counters for receive to per
context.
Profiling has shown that when mulitple contexts are being used there
is a lot of cache activity associated with these counters.
The code formerly kept these counters per port, but only provided the
interface to read per HCA. This patch converts the read of counters
to per HCA and adds the debugfs hooks to be able to read the file as a
sequence of opcodes.
Reviewed-by: Dean Luick <dean.luick@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
The current workqueue implemention has the following performance
deficiencies on QDR HCAs:
- The CQ call backs tend to run on the CPUs processing the
receive queues
- The single thread queue isn't optimal for multiple HCAs
This patch adds a dedicated per HCA bound thread to process CQ callbacks.
Reviewed-by: Ramkrishna Vepa <ramkrishna.vepa@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
This patch adds context relative numa affinity conditioned on the
module parameter numa_aware. The qib_ctxtdata has an additional
node_id member and qib_create_ctxtdata() has an addition node_id
parameter.
The allocations within the hdr queue and eager queue setup routines
now take this additional member and adjust allocations as necesary.
PSM will pass the either current numa node or the node closest to the
HCA depending on numa_aware. Verbs will always use the node closest to
the HCA.
Reviewed-by: Dean Luick <dean.luick@intel.com>
Signed-off-by: Ramkrishna Vepa <ramkrishna.vepa@intel.com>
Signed-off-by: Vinit Agnihotri <vinit.abhay.agnihotri@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
This patch adds DCA cache warming for systems that support DCA.
The code uses cpu affinity notification to react to an affinity change
from a user mode program like irqbalance and (re-)program the chip
accordingly. This notification avoids reading the current cpu on every
interrupt.
Reviewed-by: Dean Luick <dean.luick@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
[ Add Kconfig dependency on SMP && GENERIC_HARDIRQS to avoid failure to
build due to undefined struct irq_affinity_notify. - Roland ]
Signed-off-by: Roland Dreier <roland@purestorage.com>
Covers the rest of the uses of pci error handler.
Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Commit 36a8f01cd2 ("IB/qib: Add congestion control agent
implementation") tries to store the value 1984 in a u8, which leads to
truncation. Fix this by making the member big enough.
This bug was detected by a smatch warning.
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Ramkrishna Vepa <ramkrishna.vepa@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Add a congestion control agent in the driver that handles gets and
sets from the congestion control manager in the fabric for the
Performance Scale Messaging (PSM) library.
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Profiling has shown that sdma_lock is proving a bottleneck for
performance. The situations include:
- RDMA reads when krcvqs > 1
- post sends from multiple threads
For RDMA read the current global qib_wq mechanism runs on all CPUs
and contends for the sdma_lock when multiple RMDA read requests are
fielded on differenct CPUs. For post sends, the direct call to
qib_do_send() from multiple threads causes the contention.
Since the sdma mechanism is per port, this fix converts the existing
workqueue to a per port single thread workqueue to reduce the lock
contention in the RDMA read case, and for any other case where the QP
is scheduled via the workqueue mechanism from more than 1 CPU.
For the post send case, This patch modifies the post send code to test
for a non empty sdma engine. If the sdma is not idle the (now single
thread) workqueue will be used to trigger the send engine instead of
the direct call to qib_do_send().
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
This patch reorganizes the QP and devdata files to be more cache line aware.
qib_qp fields in particular are split into read-mostly, send, and receive fields.
qib_devdata fields are split into read-mostly and read/write fields
Testing has show that bidirectional tests improve by as much as 100%
with this patch.
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
This patch optimizes pio buffer allocation in the kernel.
For qib, kernel pio buffers are used for sending acks. The code to
allocate the buffer would always start at 0 until it found a buffer.
This means that an average of 64 comparisions were done on each
allocate, since the busy bit won't be cleared until the bits are
refreshed when buffers are exhausted.
This patch adds two new fields in the devdata struct, last_pio and
min_kernel_pio. last_pio is the last buffer that was allocated.
min_kernel_pio is the lowest potential available buffer.
min_kernel_pio is modifed as contexts are allocated and deallocted.
Reviewed-by: Ramkrishna Vepa <ramkrishna.vepa@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Call irq_set_affinity_hint() to give userspace programs such as
irqbalance the information to be able to distribute qib interrupts
appropriately.
The logic allocates all non-receive interrupts to the first CPU local
to the HCA. Receive interrupts are allocated round robin starting
with the second CPU local to the HCA with potential wrap back to the
second CPU.
This patch also adds a refinement to the name registered for MSI-X
interrupts so that user level scripts can determine the device
associated with the IRQs when there are multiple HCAs with a
potentially different set of local CPUs.
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@qlogic.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
The heavy weight spinlock in qib_lookup_qpn() is replaced with RCU.
The hash list itself is now accessed via jhash functions instead of mod.
The changes should benefit multiple receive contexts in different
processors by not contending for the lock just to read the hash
structures.
The patch also adds a lookaside_qp (pointer) and a lookaside_qpn in
the context. The interrupt handler will test the current packet's qpn
against lookaside_qpn if the lookaside_qp pointer is non-NULL. The
pointer is NULL'ed when the interrupt handler exits.
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@qlogic.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
The context init now saves a shift from rcvegrbufs_perchunk
rcvegrbufs_perchunk_shift using ilog2. A BUG_ON() protects the
power of 2 assumption.
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@qlogic.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
The code that was recently introduced to report the number
of free contexts is flawed for multiple HCAs:
/* Return the number of free user ports (contexts) available. */
return scnprintf(buf, PAGE_SIZE, "%u\n", dd->cfgctxts -
dd->first_user_ctxt - (u32)qib_stats.sps_ctxts);
The qib_stats is global to the module, not per HCA, so the code is broken
for multiple HCAs.
This patch adds a qib_devdata field, freectxts, that reflects the free
contexts for this HCA.
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@qlogic.com>
Reviewed-by: Ram Vepa <ram.vepa@qlogic.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
With ib_qib options:
options ib_qib krcvqs=1 pcie_caps=0x51 rcvhdrcnt=4096 singleport=1 ibmtu=4
a run of ib_write_bw -a yields the following:
------------------------------------------------------------------
#bytes #iterations BW peak[MB/sec] BW average[MB/sec]
1048576 5000 2910.64 229.80
------------------------------------------------------------------
The top cpu use in a profile is:
CPU: Intel Architectural Perfmon, speed 2400.15 MHz (estimated)
Counted CPU_CLK_UNHALTED events (Clock cycles when not halted) with a unit mask
of 0x00 (No unit mask) count 1002300
Counted LLC_MISSES events (Last level cache demand requests from this core that
missed the LLC) with a unit mask of 0x41 (No unit mask) count 10000
samples % samples % app name symbol name
15237 29.2642 964 17.1195 ib_qib.ko qib_7322intr
12320 23.6618 1040 18.4692 ib_qib.ko handle_7322_errors
4106 7.8860 0 0 vmlinux vsnprintf
Analysis of the stats, profile, the code, and the annotated profile indicate:
- All of the overflow interrupts (one per packet overflow) are
serviced on CPU0 with no mitigation on the frequency.
- All of the receive interrupts are being serviced by CPU0. (That is
the way truescale.cmds statically allocates the kctx IRQs to CPU)
- The code is spending all of its time servicing QIB_I_C_ERROR
RcvEgrFullErr interrupts on CPU0, starving the packet receive
processing.
- The decode_err routine is very inefficient, using a printf variant
to format a "%s" and continues to loop when the errs mask has been
cleared.
- Both qib_7322intr and handle_7322_errors read pci registers, which
is very inefficient.
The fix does the following:
- Adds a tasklet to service QIB_I_C_ERROR
- Replaces the very inefficient scnprintf() with a memcpy(). A field
is added to qib_hwerror_msgs to save the sizeof("string") at
compile time so that a strlen is not needed during err_decode().
- The most frequent errors (Overflows) are serviced first to exit the
loop as early as possible.
- The loop now exits as soon as the errs mask is clear rather than
fruitlessly looping through the msp array.
With this fix the performance changes to:
------------------------------------------------------------------
#bytes #iterations BW peak[MB/sec] BW average[MB/sec]
1048576 5000 2990.64 2941.35
------------------------------------------------------------------
During testing of the error handling overflow patch, it was determined
that some CPU's were slower when servicing both overflow and receive
interrupts on CPU0 with different MSI interrupt vectors.
This patch adds an option (krcvq01_no_msi) to not use a dedicated MSI
interrupt for kctx's < 2 and to service them on the default interrupt.
For some CPUs, the cost of the interrupt enter/exit is more costly
than then the additional PCI read in the default handler.
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@qlogic.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
For SusieQ we need to write to the interrupt timer register before
updating the header queue head with interrupt count. This is to
ensure that the timer is enabled properly and a receive available
interrupt is delivered. Otherwise this interrupt can be lost if the
receiver header/eager queues are full before the timer is enabled.
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@qlogic.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Noticed this odd looking thing in dmesg:
ib_qib 0000:02:00.0: <3>ib_qib: Unable to enable pcie error reporting: -5
which is due to a bad use of dev_info.
Signed-off-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Acked-by: Ralph Campbell <ralph.campbell@qlogic.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Fix build failure on sparc64 which is missing the include of
<linux/slab.h> via <asm/pci.h> that x86, powerpc, ia64, etc. have.
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Rather than use a variable size array allocation on the stack,
define a constant for the maximum array size possible.
Signed-off-by: Ralph Campbell <ralph.campbell@qlogic.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Don't set write combining via PAT on the VL15 buffers to avoid a rare
problem with unaligned writes from interrupt-flushed store buffers.
Signed-off-by: Dave Olson <dave.olson@qlogic.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Add a low-level IB driver for QLogic PCIe adapters.
Signed-off-by: Ralph Campbell <ralph.campbell@qlogic.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
