[Qemu-devel] [PATCH v9 00/20] Invert Endian bit in SPARCv9 MMU TTE

[Tony Nguyen <tony.nguyen.git@mail.com>]

From: Tony Nguyen <tony.nguyen@bt.com>

This patchset implements the IE (Invert Endian) bit in SPARCv9 MMU TTE.

It is an attempt of the instructions outlined by Richard Henderson to Mark
Cave-Ayland.

Tested with OpenBSD on sun4u. Solaris 10 is my actual goal, but unfortunately a
separate keyboard issue remains in the way.

On 01/11/17 19:15, Mark Cave-Ayland wrote:

>On 15/08/17 19:10, Richard Henderson wrote:
>
>> [CC Peter re MemTxAttrs below]
>>
>> On 08/15/2017 09:38 AM, Mark Cave-Ayland wrote:
>>> Working through an incorrect endian issue on qemu-system-sparc64, it has
>>> become apparent that at least one OS makes use of the IE (Invert Endian)
>>> bit in the SPARCv9 MMU TTE to map PCI memory space without the
>>> programmer having to manually endian-swap accesses.
>>>
>>> In other words, to quote the UltraSPARC specification: "if this bit is
>>> set, accesses to the associated page are processed with inverse
>>> endianness from what is specified by the instruction (big-for-little and
>>> little-for-big)".

A good explanation by Mark why the IE bit is required.

>>>
>>> Looking through various bits of code, I'm trying to get a feel for the
>>> best way to implement this in an efficient manner. From what I can see
>>> this could be solved using an additional MMU index, however I'm not
>>> overly familiar with the memory and softmmu subsystems.
>>
>> No, it can't be solved with an MMU index.
>>
>>> Can anyone point me in the right direction as to what would be the best
>>> way to implement this feature within QEMU?
>>
>> It's definitely tricky.
>>
>> We definitely need some TLB_FLAGS_MASK bit set so that we're forced through
>> the
>> memory slow path.  There is no other way to bypass the endianness that we've
>> already encoded from the target instruction.
>>
>> Given the tlb_set_page_with_attrs interface, I would think that we need a new
>> bit in MemTxAttrs, so that the target/sparc tlb_fill (and subroutines) can
>> pass
>> along the TTE bit for the given page.
>>
>> We have an existing problem in softmmu_template.h,
>>
>>     /* ??? Note that the io helpers always read data in the target
>>        byte ordering.  We should push the LE/BE request down into io.  */
>>     res = glue(io_read, SUFFIX)(env, mmu_idx, index, addr, retaddr);
>>     res = TGT_BE(res);
>>
>> We do not want to add a third(!) byte swap along the i/o path.  We need to
>> collapse the two that we have already before considering this one.
>>
>> This probably takes the form of:
>>
>> (1) Replacing the "int size" argument with "TCGMemOp memop" for
>>       a) io_{read,write}x in accel/tcg/cputlb.c,
>>       b) memory_region_dispatch_{read,write} in memory.c,
>>       c) adjust_endianness in memory.c.
>>     This carries size+sign+endianness down to the next level.
>>
>> (2) In memory.c, adjust_endianness,
>>
>>      if (memory_region_wrong_endianness(mr)) {
>> -        switch (size) {
>> +        memop ^= MO_BSWAP;
>> +    }
>> +    if (memop & MO_BSWAP) {
>>
>>     For extra credit, re-arrange memory_region_wrong_endianness
>>     to something more explicit -- "wrong" isn't helpful.
>
>Finally I've had a bit of spare time to experiment with this approach,
>and from what I can see there are currently 2 issues:
>
>
>1) Using TCGMemOp in memory.c means it is no longer accelerator agnostic
>
>For the moment I've defined a separate MemOp in memory.h and provided a
>mapping function in io_{read,write}x to map from TCGMemOp to MemOp and
>then pass that into memory_region_dispatch_{read,write}.
>
>Other than not referencing TCGMemOp in the memory API, another reason
>for doing this was that I wasn't convinced that all the MO_ attributes
>were valid outside of TCG. I do, of course, strongly defer to other
>people's knowledge in this area though.
>
>
>2) The above changes to adjust_endianness() fail when
>memory_region_dispatch_{read,write} are called recursively
>
>Whilst booting qemu-system-sparc64 I see that
>memory_region_dispatch_{read,write} get called recursively - once via
>io_{read,write}x and then again via flatview_read_continue() in exec.c.
>
>The net effect of this is that we perform the bswap correctly at the
>tail of the recursion, but then as we travel back up the stack we hit
>memory_region_dispatch_{read,write} once again causing a second bswap
>which means the value is returned with the incorrect endian again.
>
>
>My understanding from your softmmu_template.h comment above is that the
>memory API should do the endian swapping internally allowing the removal
>of the final TGT_BE/TGT_LE applied to the result, or did I get this wrong?
>
>> (3) In tlb_set_page_with_attrs, notice attrs.byte_swap and set
>>     a new TLB_FORCE_SLOW bit within TLB_FLAGS_MASK.
>>
>> (4) In io_{read,write}x, if iotlbentry->attrs.byte_swap is set,
>>     then memop ^= MO_BSWAP.

Thanks all for the feedback. Learnt a lot =)

v2:
- Moved size+sign+endianness attributes from TCGMemOp into MemOp.
  In v1 TCGMemOp was re-purposed entirely into MemOp.
- Replaced MemOp MO_{8|16|32|64} with TCGMemOp MO_{UB|UW|UL|UQ} alias.
  This is to avoid warnings on comparing and coercing different enums.
- Renamed get_memop to get_tcgmemop for clarity.
- MEMOP is now SIZE_MEMOP, which is just ctzl(size).
- Split patch 3/4 so one memory_region_dispatch_{read|write} interface
  is converted per patch.
- Do not reuse TLB_RECHECK, use new TLB_FORCE_SLOW instead.
- Split patch 4/4 so adding the MemTxAddrs parameters and converting
  tlb_set_page() to tlb_set_page_with_attrs() is separate from usage.
- CC'd maintainers.

v3:
- Like v1, the entire TCGMemOp enum is now MemOp.
- MemOp target dependant attributes are conditional upon NEED_CPU_H

v4:
- Added Paolo Bonzini as include/exec/memop.h maintainer

v5:
- Improved commit messages to clarify how interface to access
  MemoryRegion will be converted from "unsigned size" to "MemOp op".
- Moved cpu_transaction_failed() MemOp conversion from patch #11 to #9
  to make review easier.

v6:
- Improved commit messages.
- Include as patch #1 an earlier posted TARGET_ALIGNED_ONLY configure patch.
- Typeless macro SIZE_MEMOP is now inline.
- size_memop now includes CONFIG_DEBUG_TCG code.
- size_memop now also encodes endianness via MO_TE.
- Reverted size_memop operand "unsigned long" back to "unsigned".
- Second pass of size_memop to replace no-op place holder with MO_{8|16|32|64}.
- Delay memory_region_dispatch_{read,write} operand conversion until no-op
  size_memop is implemented so we have proper typing at all points in between.
- Fixed bug where not all memory_region_dispatch_{read,write} callers where
  encoding endianness into the MemOp operand, see patch #20.
- Fixed bug where not all memory_region_dispatch_{read,write} callers were
  collapsing their byte swap into adjust_endianness, see patch #20 and #22.
- Split byte swap collapsing patch (v5 #11) into #21 and #22.
- Corrected non-common *-common-obj to *-obj.
- Replaced enum device_endian with MemOp to simplify endianness checks. A
  straight forward sed but touched *alot* of files. See patch #16 and #17.
- Deleted enum device_endian.
- Deleted DEVICE_HOST_ENDIAN definition.
- Generalized the description of introduced MemTxAttrs attribute byte_swap.

v7:
- Fixed bug where size_memop was implicitly encoding MO_TE. Endianness,
  {MO_TE|MO_BE|MO_LE}, is now explicitly encoded by MemoryRegion accessors.
- While a no-op, size_memop return type remains an unsigned.
- Use '= 0' short hand instead of macro logic to declare host endianness.
- With new set of constant arguments, sanity checked the compiler is still
  folding away tests in cputlb.c
- Re-declared many native endian devices as little or big endian. This is why
  v7 has +16 patches.

v8:
- Split out re-declaration of native endian devices as little or big endian.
  The clean up approach was too naive, and touched too many subsystem.
- Introduced devend_memop converter which the clean up was trying to avoid.
- Collapsed bswap in flatview_{read|write}_continue into adjust_endianness.
- Fixed pci{stg|lg}_service_call to access MemoryRegion with MO_BE not MO_LE.
- Fixed memory_region_{add|del}_eventfd to adjust_endianness with target,
  not host, endianness.
- Used git send-email... apologies to all for the HTML!

v9:
- Rebase on master and test with git am... again apologies to all, thanks for
  the patience =)

Tony Nguyen (20):
  tcg: TCGMemOp is now accelerator independent MemOp
  memory: Introduce size_memop
  target/mips: Access MemoryRegion with MemOp
  hw/s390x: Access MemoryRegion with MemOp
  hw/intc/armv7m_nic: Access MemoryRegion with MemOp
  hw/virtio: Access MemoryRegion with MemOp
  hw/vfio: Access MemoryRegion with MemOp
  exec: Access MemoryRegion with MemOp
  cputlb: Access MemoryRegion with MemOp
  memory: Access MemoryRegion with MemOp
  hw/s390x: Hard code size with MO_{8|16|32|64}
  target/mips: Hard code size with MO_{8|16|32|64}
  exec: Hard code size with MO_{8|16|32|64}
  memory: Access MemoryRegion with endianness
  cputlb: Replace size and endian operands for MemOp
  memory: Single byte swap along the I/O path
  cpu: TLB_FLAGS_MASK bit to force memory slow path
  cputlb: Byte swap memory transaction attribute
  target/sparc: Add TLB entry with attributes
  target/sparc: sun4u Invert Endian TTE bit

 MAINTAINERS                             |   1 +
 accel/tcg/cputlb.c                      | 197 ++++++++++--------------
 exec.c                                  |  10 +-
 hw/intc/armv7m_nvic.c                   |  13 +-
 hw/s390x/s390-pci-inst.c                |  11 +-
 hw/vfio/pci-quirks.c                    |   7 +-
 hw/virtio/virtio-pci.c                  |  15 +-
 include/exec/cpu-all.h                  |  10 +-
 include/exec/memattrs.h                 |   2 +
 include/exec/memop.h                    | 134 ++++++++++++++++
 include/exec/memory.h                   |  12 +-
 memory.c                                |  58 ++++---
 memory_ldst.inc.c                       |  81 ++--------
 target/alpha/translate.c                |   2 +-
 target/arm/translate-a64.c              |  48 +++---
 target/arm/translate-a64.h              |   2 +-
 target/arm/translate-sve.c              |   2 +-
 target/arm/translate.c                  |  32 ++--
 target/arm/translate.h                  |   2 +-
 target/hppa/translate.c                 |  14 +-
 target/i386/translate.c                 | 132 ++++++++--------
 target/m68k/translate.c                 |   2 +-
 target/microblaze/translate.c           |   4 +-
 target/mips/op_helper.c                 |   5 +-
 target/mips/translate.c                 |   8 +-
 target/openrisc/translate.c             |   4 +-
 target/ppc/translate.c                  |  12 +-
 target/riscv/insn_trans/trans_rva.inc.c |   8 +-
 target/riscv/insn_trans/trans_rvi.inc.c |   4 +-
 target/s390x/translate.c                |   6 +-
 target/s390x/translate_vx.inc.c         |  10 +-
 target/sparc/cpu.h                      |   2 +
 target/sparc/mmu_helper.c               |  40 +++--
 target/sparc/translate.c                |  14 +-
 target/tilegx/translate.c               |  10 +-
 target/tricore/translate.c              |   8 +-
 tcg/README                              |   2 +-
 tcg/aarch64/tcg-target.inc.c            |  26 ++--
 tcg/arm/tcg-target.inc.c                |  26 ++--
 tcg/i386/tcg-target.inc.c               |  24 +--
 tcg/mips/tcg-target.inc.c               |  16 +-
 tcg/optimize.c                          |   2 +-
 tcg/ppc/tcg-target.inc.c                |  12 +-
 tcg/riscv/tcg-target.inc.c              |  20 +--
 tcg/s390/tcg-target.inc.c               |  14 +-
 tcg/sparc/tcg-target.inc.c              |   6 +-
 tcg/tcg-op.c                            |  38 ++---
 tcg/tcg-op.h                            |  86 +++++------
 tcg/tcg.c                               |   2 +-
 tcg/tcg.h                               | 101 +-----------
 trace/mem-internal.h                    |   4 +-
 trace/mem.h                             |   4 +-
 52 files changed, 662 insertions(+), 643 deletions(-)
 create mode 100644 include/exec/memop.h

--
2.23.0