target/linux/layerscape/patches-5.4/701-net-0155-soc-fsl-dpio-aligned-access-of-qbman-cacheable-regio.patch

   1 From 353725d455cc27aa2008456c2711e7fae2b8a631 Mon Sep 17 00:00:00 2001
   2 From: Bharat Bhushan <Bharat.Bhushan@nxp.com>
   3 Date: Wed, 7 Mar 2018 21:35:37 +0530
   4 Subject: [PATCH] soc: fsl: dpio: aligned access of qbman cacheable region
   5
   6 Alignment requirement on ARM is lenient (In Linux) for regions
   7 mapped as "Memory Type" but have very strict policy for regions
   8 mapped as "Device Type". Unaligned access to regions mapped
   9 as "Device Type" will always result to unaligned fault.
  10
  11 DPIO driver have un-aligned access to QBman cacheable region
  12 and the Linux driver maps the region as "Memory Type". On Host
  13 Linux this works because MMU Stage-1 configured by driver allows
  14 unaligned access. In Virtual Machine cases, final region mapping type
  15 is governed by combination of Stage-1 and Stage-2 MMU mapping.
  16
  17 Linux driver in VM controls maps the region as "Memory Type" in
  18 Stage-1 MMU while Stage-2 is controlled by KVM. And current KVM
  19 implementation does not allow device region to be mapped as
  20 "Memory Type". Till we have a working/upstream-able solution
  21 for Virtual Machine, we need to change un-aligned access in DPIO
  22 driver to be aligned
  23
  24 While we reached to this point as we observed below alignment
  25 exception in Virtual Machine when accessing qbman cacheable region.
  26
  27   kvm [2347]: Unsupported FSC: EC=0x24 xFSC=0x21
  28   ESR_EL2=0x92000061
  29   error: kvm run failed Bad address
  30   PC=ffff000008398e78  SP=ffff800009bcb540
  31   X00=ffff000008041000 X01=ffff800009bcb580 X02=ffff800009bcb650
  32   X03=0000000000000180
  33   X04=ffff000008041001 X05=ffff800009bcb581 X06=0200000000000000
  34   X07=0000000000000000
  35   X08=0000000000000000 X09=ffff000008041000 X10=0000000000000001
  36   X11=0000000000de6cb0
  37   X12=00000000fa83b2da X13=0000000000000001 X14=000000007f605ec8
  38   X15=00000000e26f5d5e
  39   X16=000000008521af1e X17=000000001076277e X18=ffff800009bcb5c0
  40   X19=ffff800079da2b00
  41   X20=ffff800009bcb650 X21=0000000000000002 X22=0000000000000000
  42   X23=0000000000000000
  43   X24=0000000000000000 X25=ffff8000099e7440 X26=ffff000008da6000
  44   X27=ffff000008e7f000
  45   X28=00000000499e7440 X29=ffff800009bcb540 X30=ffff00000839a160
  46   PSTATE=20000145 --C- EL1h
  47
  48 Signed-off-by: Bharat Bhushan <Bharat.Bhushan@nxp.com>
  49 ---
  50  drivers/soc/fsl/dpio/qbman-portal.c | 12 +++++++++++-
  51  1 file changed, 11 insertions(+), 1 deletion(-)
  52
  53 --- a/drivers/soc/fsl/dpio/qbman-portal.c
  54 +++ b/drivers/soc/fsl/dpio/qbman-portal.c
  55 @@ -515,7 +515,17 @@ int qbman_swp_enqueue(struct qbman_swp *
  56                 return -EBUSY;
  57
  58         p = qbman_get_cmd(s, QBMAN_CENA_SWP_EQCR(EQAR_IDX(eqar)));
  59 -       memcpy(&p->dca, &d->dca, 31);
  60 +       /* This is mapped as DEVICE type memory, writes are
  61 +        * with address alignment:
  62 +        * desc.dca address alignment = 1
  63 +        * desc.seqnum address alignment = 2
  64 +        * desc.orpid address alignment = 4
  65 +        * desc.tgtid address alignment = 8
  66 +        */
  67 +       p->dca = d->dca;
  68 +       p->seqnum = d->seqnum;
  69 +       p->orpid = d->orpid;
  70 +       memcpy(&p->tgtid, &d->tgtid, 24);
  71         memcpy(&p->fd, fd, sizeof(*fd));
  72
  73         if ((s->desc->qman_version & QMAN_REV_MASK) < QMAN_REV_5000) {