Skip to content

K
Kernel

openeuler / Kernel
1 年多 前同步成功

通知 8
Star 0
Fork 0
K
Kernel
切换分支/标签
  1. 30 11月, 2016 4 次提交
    • C
      SUNRPC: Proper metric accounting when RPC is not transmitted · ae09531d
      由 Chuck Lever 提交于 
      I noticed recently that during an xfstests on a krb5i mount, the
retransmit count for certain operations had gone negative, and the
backlog value became unreasonably large. I recall that Andy has
pointed this out to me in the past.

When call_refresh fails to find a valid credential for an RPC, the
RPC exits immediately without sending anything on the wire. This
leaves rq_ntrans, rq_xtime, and rq_rtt set to zero.

The solution for om_queue is to not add the to RPC's running backlog
queue total whenever rq_xtime is zero.

For om_ntrans, it's a bit more difficult. A zero rq_ntrans causes
om_ops to become larger than om_ntrans. The design of the RPC
metrics API assumes that ntrans will always be equal to or larger
than the ops count. The result is that when an RPC fails to find
credentials, the RPC operation's reported retransmit count, which is
computed in user space as the difference between ops and ntrans,
goes negative.

Ideally the kernel API should report a separate retransmit and
"exited before initial transmission" metric, so that user space can
sort out the difference properly.

To avoid kernel API changes and changes to the way rq_ntrans is used
when performing transport locking, account for untransmitted RPCs
so that om_ntrans keeps up with om_ops: always add one or more to
om_ntrans.
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
Signed-off-by: NAnna Schumaker <Anna.Schumaker@Netapp.com>
      ae09531d
    • C
      xprtrdma: Support for SG_GAP devices · 5e9fc6a0
      由 Chuck Lever 提交于 
      Some devices (such as the Mellanox CX-4) can register, under a
single R_key, a set of memory regions that are not contiguous. When
this is done, all the segments in a Reply list, say, can then be
invalidated in a single LocalInv Work Request (or via Remote
Invalidation, which can invalidate exactly one R_key when completing
a Receive).

This means a single FastReg WR is used to register, and one or zero
LocalInv WRs can invalidate, the memory involved with RDMA transfers
on behalf of an RPC.

In addition, xprtrdma constructs some Reply chunks from three or
more segments. By registering them with SG_GAP, only one segment
is needed for the Reply chunk, allowing the whole chunk to be
invalidated remotely.
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
Signed-off-by: NAnna Schumaker <Anna.Schumaker@Netapp.com>
      5e9fc6a0
    • C
      xprtrdma: Make FRWR send queue entry accounting more accurate · 8d38de65
      由 Chuck Lever 提交于 
      Verbs providers may perform house-keeping on the Send Queue during
each signaled send completion. It is necessary therefore for a verbs
consumer (like xprtrdma) to occasionally force a signaled send
completion if it runs unsignaled most of the time.

xprtrdma does not require signaled completions for Send or FastReg
Work Requests, but does signal some LocalInv Work Requests. To
ensure that Send Queue house-keeping can run before the Send Queue
is more than half-consumed, xprtrdma forces a signaled completion
on occasion by counting the number of Send Queue Entries it
consumes. It currently does this by counting each ib_post_send as
one Entry.

Commit c9918ff5 ("xprtrdma: Add ro_unmap_sync method for FRWR")
introduced the ability for frwr_op_unmap_sync to post more than one
Work Request with a single post_send. Thus the underlying assumption
of one Send Queue Entry per ib_post_send is no longer true.

Also, FastReg Work Requests are currently never signaled. They
should be signaled once in a while, just as Send is, to keep the
accounting of consumed SQEs accurate.

While we're here, convert the CQCOUNT macros to the currently
preferred kernel coding style, which is inline functions.

Fixes: c9918ff5 ("xprtrdma: Add ro_unmap_sync method for FRWR")
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
Signed-off-by: NAnna Schumaker <Anna.Schumaker@Netapp.com>
      8d38de65
    • C
      xprtrdma: Cap size of callback buffer resources · 62aee0e3
      由 Chuck Lever 提交于 
      When the inline threshold size is set to large values (say, 32KB)
any NFSv4.1 CB request from the server gets a reply with status
NFS4ERR_RESOURCE.

Looks like there are some upper layer assumptions about the maximum
size of a reply (for example, in process_op). Cap the size of the
NFSv4 client's reply resources at a page.
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
Signed-off-by: NAnna Schumaker <Anna.Schumaker@Netapp.com>
      62aee0e3
  3. 28 11月, 2016 3 次提交
  5. 27 11月, 2016 8 次提交
  7. 26 11月, 2016 22 次提交
  9. 25 11月, 2016 3 次提交
    • J
      parisc: Also flush data TLB in flush_icache_page_asm · 5035b230
      由 John David Anglin 提交于 
      This is the second issue I noticed in reviewing the parisc TLB code.

The fic instruction may use either the instruction or data TLB in
flushing the instruction cache.  Thus, on machines with a split TLB, we
should also flush the data TLB after setting up the temporary alias
registers.

Although this has no functional impact, I changed the pdtlb and pitlb
instructions to consistently use the index register %r0.  These
instructions do not support integer displacements.

Tested on rp3440 and c8000.
Signed-off-by: NJohn David Anglin  <dave.anglin@bell.net>
Cc: <stable@vger.kernel.org> # v3.16+
Signed-off-by: NHelge Deller <deller@gmx.de>
      5035b230
    • J
      parisc: Fix race in pci-dma.c · c0452fb9
      由 John David Anglin 提交于 
      We are still troubled by occasional random segmentation faults and
memory memory corruption on SMP machines.  The causes quite a few
package builds to fail on the Debian buildd machines for parisc.  When
gcc-6 failed to build three times in a row, I looked again at the TLB
related code.  I found a couple of issues.  This is the first.

In general, we need to ensure page table updates and corresponding TLB
purges are atomic.  The attached patch fixes an instance in pci-dma.c
where the page table update was not guarded by the TLB lock.

Tested on rp3440 and c8000.  So far, no further random segmentation
faults have been observed.
Signed-off-by: NJohn David Anglin  <dave.anglin@bell.net>
Cc: <stable@vger.kernel.org> # v3.16+
Signed-off-by: NHelge Deller <deller@gmx.de>
      c0452fb9
    • H
      parisc: Switch to generic sched_clock implementation · 43b1f6ab
      由 Helge Deller 提交于 
      Drop the open-coded sched_clock() function and replace it by the provided
GENERIC_SCHED_CLOCK implementation.  We have seen quite some hung tasks in the
past, which seem to be fixed by this patch.
Signed-off-by: NHelge Deller <deller@gmx.de>
Cc: <stable@vger.kernel.org> # v4.7+
Signed-off-by: NHelge Deller <deller@gmx.de>
      43b1f6ab