We now expose qemu_announce_self through QMP and HMP.  Add a test
with some very basic packet validation (make sure we get a RARP).
Signed-off-by: NVlad Yasevich <vyasevic@redhat.com>
Signed-off-by: NDr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NJason Wang <jasowang@redhat.com>

Add an HMP command to trigger self annocements.
Unlike the QMP command (which takes a set of parameters), the HMP
command reuses the set of parameters used for migration.
Signend-off-by: NVladislav Yasevich <vyasevic@redhat.com>
Signed-off-by: NDr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NJason Wang <jasowang@redhat.com>

Add a qmp command that can trigger guest announcements.

It uses its own announce-timer instance, and parameters
passed to it explicitly in the command.

Like most qmp commands, it's in the main thread/bql, so
there's no racing with any outstanding timer.

Based on work of Germano Veit Michel <germano@redhat.com> and
                 Vladislav Yasevich <vyasevic@redhat.com>
Signed-off-by: NDr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NJason Wang <jasowang@redhat.com>

Expose the virtio-net self announcement capability and allow
qemu_announce_self() to call it.

These announces are caused by something external (i.e. the
announce-self command); they won't trigger if the migration
counter is triggering announces at the same time.
Signed-off-by: NVladislav Yasevich <vyasevic@redhat.com>
Signed-off-by: NDr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NJason Wang <jasowang@redhat.com>

Some network devices have a capability to do self announcements
(ex: virtio-net).  Add infrastructure that would allow devices
to expose this ability.
Signed-off-by: NVladislav Yasevich <vyasevic@redhat.com>
Signed-off-by: NDr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NJason Wang <jasowang@redhat.com>

Switch the announcements to using the new announce timer.
Move the code that does it to announce.c rather than savevm
because it really has nothing to do with the actual migration.

Migration starts the announce from bh's and so they're all
in the main thread/bql, and so there's never any racing with
the timers themselves.
Signed-off-by: NDr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NJason Wang <jasowang@redhat.com>

Switch virtio's self announcement to use the AnnounceTimer.
It keeps it's own AnnounceTimer (per device), and starts running it
using a migration post-load and a virtual clock; that way the
announce happens once the guest is actually running.
The timer uses the migration parameters to set the timing of
the repeats.

Based on earlier patches by myself and
 Vladislav Yasevich <vyasevic@redhat.com>
Signed-off-by: NDr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NJason Wang <jasowang@redhat.com>

Add migration parameters that control RARP/GARP announcement timeouts.

Based on earlier patches by myself and
  Vladislav Yasevich <vyasevic@redhat.com>
Signed-off-by: NDr. David Alan Gilbert <dgilbert@redhat.com>
Acked-by: NMarkus Armbruster <armbru@redhat.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NJason Wang <jasowang@redhat.com>

The 'announce timer' will be used by migration, and explicit
requests for qemu to perform network announces.

Based on the work by Germano Veit Michel <germano@redhat.com>
 and Vlad Yasevich <vyasevic@redhat.com>
Signed-off-by: NDr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NJason Wang <jasowang@redhat.com>

Changes:
  - Save CPU cycles by computing the return value while scanning the
    input iovec, rather than calling iov_size() at the end.
  - Remove check for s->tx != NULL, because it cannot happen.
  - Cache ring->tail in a local variable and use it to check for
    space in the TX ring. The use of nm_ring_empty() was invalid,
    because nobody is updating ring->cur and ring->head at that point.
  - In case we run out of netmap slots in the middle of a packet,
    move the wake-up point by advancing ring->cur, but do not
    expose the incomplete packet (i.e., by updating also ring->head).
Signed-off-by: NVincenzo Maffione <v.maffione@gmail.com>
Signed-off-by: NJason Wang <jasowang@redhat.com>

Improve code reuse by implementing netmap_receive() with a call
to netmap_receive_iov().
Signed-off-by: NVincenzo Maffione <v.maffione@gmail.com>
Signed-off-by: NJason Wang <jasowang@redhat.com>

This change improves the handling of incomplete multi-slot packets
(e.g. with the NS_MOREFRAG set), by advancing ring->head only on
complete packets. The ring->cur pointer is advanced in any case in
order to acknowledge the kernel and move the wake-up point (thus
avoiding repeated wake-ups).
Also don't be verbose when incomplete packets are found.
Signed-off-by: NVincenzo Maffione <v.maffione@gmail.com>
Signed-off-by: NJason Wang <jasowang@redhat.com>

Fix duplicated code:
https://bugs.launchpad.net/qemu/+bug/1811499Reviewed-by: NThomas Huth <thuth@redhat.com>
Reviewed-by: NPhilippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: NZhang Chen <chen.zhang@intel.com>
Signed-off-by: NJason Wang <jasowang@redhat.com>

cocoa tree:
 * Support OSX Mojave by ensuring that we always make Cocoa UI
   function calls from the main thread, never from any other QEMU
   thread. This was previously mostly harmless, but on Mojave
   it will cause OSX to terminate the QEMU process.

# gpg: Signature made Mon 04 Mar 2019 16:48:57 GMT
# gpg:                using RSA key E1A5C593CD419DE28E8315CF3C2525ED14360CDE
# gpg:                issuer "peter.maydell@linaro.org"
# gpg: Good signature from "Peter Maydell <peter.maydell@linaro.org>" [ultimate]
# gpg:                 aka "Peter Maydell <pmaydell@gmail.com>" [ultimate]
# gpg:                 aka "Peter Maydell <pmaydell@chiark.greenend.org.uk>" [ultimate]
# Primary key fingerprint: E1A5 C593 CD41 9DE2 8E83  15CF 3C25 25ED 1436 0CDE

* remotes/pmaydell/tags/pull-cocoa-20190304:
  ui/cocoa: Perform UI operations only on the main thread
  ui/cocoa: Subclass NSApplication so we can implement sendEvent
  ui/cocoa: Don't call NSApp sendEvent directly from handleEvent
  ui/cocoa: Move console/device menu creation code up in file
  ui/cocoa: Factor out initial menu creation
  ui/cocoa: Use the pixman image directly in switchSurface
  ui/cocoa: Ensure we have the iothread lock when calling into QEMU
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

The OSX Mojave release is more picky about enforcing the Cocoa API
restriction that only the main thread may perform UI calls. To
accommodate this we need to restructure the Cocoa code:
 * the special OSX main() creates a second thread and uses
   that to call the vl.c qemu_main(); the original main
   thread goes into the OSX event loop
 * the refresh, switch and update callbacks asynchronously
   tell the main thread to do the necessary work
 * the refresh callback no longer does the "get events from the
   UI event queue and handle them" loop, since we now use
   the stock OSX event loop. Instead our NSApplication sendEvent
   method will either deal with them or pass them on to OSX

All these things have to be changed in one commit, to avoid
breaking bisection.

Note that since we use dispatch_get_main_queue(), this bumps
our minimum version requirement to OSX 10.10 Yosemite (released
in 2014, unsupported by Apple since 2017).
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>
Reviewed-by: NRoman Bolshakov <r.bolshakov@yadro.com>
Tested-by: NRoman Bolshakov <r.bolshakov@yadro.com>
Message-id: 20190225102433.22401-8-peter.maydell@linaro.org
Message-id: 20190214102816.3393-8-peter.maydell@linaro.org

When we switch away from our custom event handling, we still want to
be able to have first go at any events our application receives,
because in full-screen mode we want to send key events to the guest,
even if they would be menu item activation events. There are several
ways we could do that, but one simple approach is to subclass
NSApplication so we can implement a custom sendEvent method.
Do that, but for the moment have our sendEvent just invoke the
superclass method.
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>
Reviewed-by: NBALATON Zoltan <balaton@eik.bme.hu>
Reviewed-by: NRoman Bolshakov <r.bolshakov@yadro.com>
Tested-by: NRoman Bolshakov <r.bolshakov@yadro.com>
Message-id: 20190225102433.22401-7-peter.maydell@linaro.org
Message-id: 20190214102816.3393-7-peter.maydell@linaro.org

Currently the handleEvent method will directly call the NSApp
sendEvent method for any events that we want to let OSX deal
with. When we rearrange the event handling code, the way that
we say "let OSX have this event" is going to change. Prepare
for that by refactoring so that handleEvent returns a flag
indicating whether it consumed the event.
Suggested-by: NBALATON Zoltan <balaton@eik.bme.hu>
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>
Reviewed-by: NBALATON Zoltan <balaton@eik.bme.hu>
Reviewed-by: NRoman Bolshakov <r.bolshakov@yadro.com>
Tested-by: NRoman Bolshakov <r.bolshakov@yadro.com>
Message-id: 20190225102433.22401-6-peter.maydell@linaro.org
Message-id: 20190214102816.3393-6-peter.maydell@linaro.org

Move the console/device menu creation code functions
further up in the source file, next to the code which
creates the initial menus. We're going to want to
change the location we call these functions from in
the next patch.

This commit is a pure code move with no other changes.
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>
Reviewed-by: NBALATON Zoltan <balaton@eik.bme.hu>
Reviewed-by: NRoman Bolshakov <r.bolshakov@yadro.com>
Tested-by: NRoman Bolshakov <r.bolshakov@yadro.com>
Message-id: 20190225102433.22401-5-peter.maydell@linaro.org
Message-id: 20190214102816.3393-5-peter.maydell@linaro.org

Factor out the long code sequence in main() which creates
the initial set of menus. This will make later patches
which move initialization code around a bit clearer.
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>
Reviewed-by: NBALATON Zoltan <balaton@eik.bme.hu>
Reviewed-by: NRoman Bolshakov <r.bolshakov@yadro.com>
Tested-by: NRoman Bolshakov <r.bolshakov@yadro.com>
Message-id: 20190225102433.22401-4-peter.maydell@linaro.org
Message-id: 20190214102816.3393-4-peter.maydell@linaro.org

Currently the switchSurface method takes a DisplaySurface. We want
to change our DisplayChangeListener's dpy_gfx_switch callback
to do this work asynchronously on a different thread. The caller
of the switch callback will free the old DisplaySurface
immediately the callback returns, so to ensure that the
other thread doesn't access freed data we need to switch
to using the underlying pixman image instead. The pixman
image is reference counted, so we will be able to take
a reference to it to avoid it vanishing too early.

In this commit we only change the switchSurface method
to take a pixman image, and keep the flow of control
synchronous for now.
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>
Reviewed-by: NBALATON Zoltan <balaton@eik.bme.hu>
Reviewed-by: NRoman Bolshakov <r.bolshakov@yadro.com>
Tested-by: NRoman Bolshakov <r.bolshakov@yadro.com>
Message-id: 20190225102433.22401-3-peter.maydell@linaro.org
Message-id: 20190214102816.3393-3-peter.maydell@linaro.org

The Cocoa UI should run on the main thread; this is enforced
in OSX Mojave. In order to be able to run on the main thread,
we need to make sure we hold the iothread lock whenever we
call into various QEMU UI midlayer functions.
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>
Reviewed-by: NRoman Bolshakov <r.bolshakov@yadro.com>
Tested-by: NRoman Bolshakov <r.bolshakov@yadro.com>
Message-id: 20190225102433.22401-2-peter.maydell@linaro.org
Message-id: 20190214102816.3393-2-peter.maydell@linaro.org

Merge tpm 2029/02/25 v1

# gpg: Signature made Mon 25 Feb 2019 15:05:12 GMT
# gpg:                using RSA key 75AD65802A0B4211
# gpg: Good signature from "Stefan Berger <stefanb@linux.vnet.ibm.com>" [unknown]
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg:          There is no indication that the signature belongs to the owner.
# Primary key fingerprint: B818 B9CA DF90 89C2 D5CE  C66B 75AD 6580 2A0B 4211

* remotes/stefanberger/tags/pull-tpm-2019-02-25-1:
  tpm_tis: convert tpm_tis_show_buffer() to use trace event
  tpm_tis: fix loop that cancels any seizure by a lower locality
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

s390x updates:
- tcg: support the floating-point extension facility
- vfio-ap: support hot(un)plug of vfio-ap device
- fixes + cleanups

# gpg: Signature made Mon 04 Mar 2019 11:55:39 GMT
# gpg:                using RSA key C3D0D66DC3624FF6A8C018CEDECF6B93C6F02FAF
# gpg:                issuer "cohuck@redhat.com"
# gpg: Good signature from "Cornelia Huck <conny@cornelia-huck.de>" [unknown]
# gpg:                 aka "Cornelia Huck <huckc@linux.vnet.ibm.com>" [full]
# gpg:                 aka "Cornelia Huck <cornelia.huck@de.ibm.com>" [full]
# gpg:                 aka "Cornelia Huck <cohuck@kernel.org>" [unknown]
# gpg:                 aka "Cornelia Huck <cohuck@redhat.com>" [unknown]
# Primary key fingerprint: C3D0 D66D C362 4FF6 A8C0  18CE DECF 6B93 C6F0 2FAF

* remotes/cohuck/tags/s390x-20190304: (27 commits)
  s390x: Add floating-point extension facility to "qemu" cpu model
  s390x/tcg: Handle all rounding modes overwritten by BFP instructions
  s390x/tcg: Implement rounding mode and XxC for LOAD ROUNDED
  s390x/tcg: Implement XxC and checks for most FP instructions
  s390x/tcg: Prepare for IEEE-inexact-exception control (XxC)
  s390x/tcg: Refactor saving/restoring the bfp rounding mode
  s390x/tcg: Check for exceptions in SET BFP ROUNDING MODE
  s390x/tcg: Handle SET FPC AND LOAD FPC 3-bit BFP rounding modes
  s390x/tcg: Fix simulated-IEEE exceptions
  s390x/tcg: Refactor SET FPC AND SIGNAL handling
  s390x/tcg: Hide IEEE underflows in some scenarios
  s390x/tcg: Fix parts of IEEE exception handling
  s390x/tcg: Factor out conversion of softfloat exceptions
  s390x/tcg: Fix rounding from float128 to uint64_t/uint32_t
  s390x/tcg: Fix TEST DATA CLASS instructions
  s390x/tcg: Implement LOAD COUNT TO BLOCK BOUNDARY
  s390x/tcg: Implement LOAD LENGTHENED short HFP to long HFP
  s390x/tcg: Factor out gen_addi_and_wrap_i64() from get_address()
  s390x/tcg: Factor out vec_full_reg_offset()
  s390x/tcg: Clarify terminology in vec_reg_offset()
  ...
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

pci, pc, virtio: fixes, cleanups, tests

Lots of work on tests: BiosTablesTest UEFI app,
vhost-user testing for non-Linux hosts.
Misc cleanups and fixes all over the place
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>

# gpg: Signature made Fri 22 Feb 2019 15:51:40 GMT
# gpg:                using RSA key 281F0DB8D28D5469
# gpg: Good signature from "Michael S. Tsirkin <mst@kernel.org>" [full]
# gpg:                 aka "Michael S. Tsirkin <mst@redhat.com>" [full]
# Primary key fingerprint: 0270 606B 6F3C DF3D 0B17  0970 C350 3912 AFBE 8E67
#      Subkey fingerprint: 5D09 FD08 71C8 F85B 94CA  8A0D 281F 0DB8 D28D 5469

* remotes/mst/tags/for_upstream: (26 commits)
  pci: Sanity test minimum downstream LNKSTA
  hw/smbios: fix offset of type 3 sku field
  pci: Move NVIDIA vendor id to the rest of ids
  virtio-balloon: Safely handle BALLOON_PAGE_SIZE < host page size
  virtio-balloon: Use ram_block_discard_range() instead of raw madvise()
  virtio-balloon: Rework ballon_page() interface
  virtio-balloon: Corrections to address verification
  virtio-balloon: Remove unnecessary MADV_WILLNEED on deflate
  i386/kvm: ignore masked irqs when update msi routes
  contrib/vhost-user-blk: fix the compilation issue
  Revert "contrib/vhost-user-blk: fix the compilation issue"
  pc-dimm: use same mechanism for [get|set]_addr
  tests/data: introduce "uefi-boot-images" with the "bios-tables-test" ISOs
  tests/uefi-test-tools: add build scripts
  tests: introduce "uefi-test-tools" with the BiosTablesTest UEFI app
  roms: build the EfiRom utility from the roms/edk2 submodule
  roms: add the edk2 project as a git submodule
  vhost-user-test: create a temporary directory per TestServer
  vhost-user-test: small changes to init_hugepagefs
  vhost-user-test: create a main loop per TestServer
  ...
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

The floating-point extension facility implemented certain changes to
BFP, HFP and DFP instructions.

As we don't implement HFP/DFP, we can ignore those completely. Related
to BFP, the changes include
- SET BFP ROUNDING MODE (SRNMB) instruction
- BFP-rounding-mode field in the FPC register is changed to 3 bits
- CONVERT FROM LOGICAL instructions
- CONVERT TO LOGICAL instructions
- Changes (rounding mode + XxC) added to
-- CONVERT TO FIXED
-- CONVERT FROM FIXED
-- LOAD FP INTEGER
-- LOAD ROUNDED
-- DIVIDE TO INTEGER

For TCG, we don't implement DIVIDE TO INTEGER, and it is harder to
implement, so skip that. Also, as we don't implement PFPO, we can skip
changes to that as well. The other parts are now implemented, we can
indicate the facility.

z14 PoP mentions that "The floating-point extension facility is installed
in the z/Architecture architectural mode. When bit 37 is one, bit 42 is
also one.", meaning that the DFP (decimal-floating-point) facility also
has to be indicated. We can ignore that for now.
Signed-off-by: NDavid Hildenbrand <david@redhat.com>
Message-Id: <20190218122710.23639-16-david@redhat.com>
Reviewed-by: NThomas Huth <thuth@redhat.com>
Signed-off-by: NCornelia Huck <cohuck@redhat.com>

"round to nearest with ties away from 0" maps to float_round_ties_away.
"round to prepare for shorter precision" maps to float_round_to_odd.

As all instructions properly check for valid rounding modes in translate.c
we can add an assert. Fix one missing empty line.

Cc: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: NRichard Henderson <richard.henderson@linaro.org>
Signed-off-by: NDavid Hildenbrand <david@redhat.com>
Message-Id: <20190218122710.23639-15-david@redhat.com>
Signed-off-by: NCornelia Huck <cohuck@redhat.com>

With the floating-point extension facility, LOAD ROUNDED has
a rounding mode specification and the inexact-exception control (XxC).

Handle them just like e.g. LOAD FP INTEGER.
Reviewed-by: NRichard Henderson <richard.henderson@linaro.org>
Signed-off-by: NDavid Hildenbrand <david@redhat.com>
Message-Id: <20190218122710.23639-14-david@redhat.com>
Signed-off-by: NCornelia Huck <cohuck@redhat.com>

With the floating-point extension facility
- CONVERT FROM LOGICAL
- CONVERT TO LOGICAL
- CONVERT TO FIXED
- CONVERT FROM FIXED
- LOAD FP INTEGER
have both, a rounding mode specification and the inexact-exception control
(XxC). Other instructions will be handled separatly.

Check for valid rounding modes and forward also the XxC (via m4). To avoid
a lot of boilerplate code and changes to the helpers, combine both, the
m3 and m4 field in a combined 32 bit TCG variable. Perform checks at
a central place, taking in account if the m3 or m4 field was ignore
before the floating-point extension facility was introduced.
Signed-off-by: NDavid Hildenbrand <david@redhat.com>
Message-Id: <20190218122710.23639-13-david@redhat.com>
Reviewed-by: NRichard Henderson <richard.henderson@linaro.org>
Signed-off-by: NCornelia Huck <cohuck@redhat.com>

Some instructions allow to suppress IEEE inexact exceptions.

z14 PoP, 9-23, "Suppression of Certain IEEE Exceptions"
    IEEE-inexact-exception control (XxC): Bit 1 of
    the M4 field is the XxC bit. If XxC is zero, recogni-
    tion of IEEE-inexact exception is not suppressed;
    if XxC is one, recognition of IEEE-inexact excep-
    tion is suppressed.

Especially, handling for overflow/unerflow remains as is, inexact is
reported along

z14 PoP, 9-23, "Suppression of Certain IEEE Exceptions"
    For example, the IEEE-inexact-exception control (XxC)
    has no effect on the DXC; that is, the DXC for IEEE-
    overflow or IEEE-underflow exceptions along with the
    detail for exact, inexact and truncated, or inexact and
    incremented, is reported according to the actual con-
    dition.

Follow up patches will wire it correctly up for the applicable
instructions.
Reviewed-by: NRichard Henderson <richard.henderson@linaro.org>
Signed-off-by: NDavid Hildenbrand <david@redhat.com>
Message-Id: <20190218122710.23639-12-david@redhat.com>
Signed-off-by: NCornelia Huck <cohuck@redhat.com>

We want to reuse this in the context of vector instructions. So use
better matching names and introduce s390_restore_bfp_rounding_mode().

While at it, add proper newlines.
Reviewed-by: NRichard Henderson <richard.henderson@linaro.org>
Signed-off-by: NDavid Hildenbrand <david@redhat.com>
Message-Id: <20190218122710.23639-11-david@redhat.com>
Signed-off-by: NCornelia Huck <cohuck@redhat.com>

Let's split handling of BFP/DFP rounding mode configuration. Also,
let's not reuse the sfpc handler, use a separate handler so we can
properly check for specification exceptions for SRNMB.
Reviewed-by: NRichard Henderson <richard.henderson@linaro.org>
Signed-off-by: NDavid Hildenbrand <david@redhat.com>
Message-Id: <20190218122710.23639-10-david@redhat.com>
Signed-off-by: NCornelia Huck <cohuck@redhat.com>

We already forward the 3 bits correctly in the translation functions. We
also have to handle them properly and check for specification
exceptions.

Setting an invalid rounding mode (BFP only, all DFP rounding modes)
results in a specification exception. Setting unassigned bits in the
fpc, results in a specification exception.

This fixes LOAD FPC (AND SIGNAL), SET FPC (AND SIGNAL). Also for,
SET BFP ROUNDING MODE, 3-bit rounding mode is now explicitly checked.

Note: TCG_CALL_NO_WG is required for sfpc handler, as we now inject
exceptions.

We won't be modeling abscence of the "floating-point extension facility"
for now, not necessary as most take the facility for granted without
checking.

z14 PoP, 9-23, "LOAD FPC"
    When the floating-point extension facility is
    installed, bits 29-31 of the second operand must
    specify a valid BFP rounding mode and bits 6-7,
    14-15, 24, and 28 must be zero; otherwise, a
    specification exception is recognized.
Signed-off-by: NDavid Hildenbrand <david@redhat.com>
Message-Id: <20190218122710.23639-9-david@redhat.com>
Reviewed-by: NRichard Henderson <richard.henderson@linaro.org>
Signed-off-by: NCornelia Huck <cohuck@redhat.com>

The trap is triggered based on priority of the enabled signaling flags.
Only overflow and underflow allow a concurrent inexact exception.

z14 PoP, 9-33, Figure 9-21
Reviewed-by: NRichard Henderson <richard.henderson@linaro.org>
Signed-off-by: NDavid Hildenbrand <david@redhat.com>
Message-Id: <20190218122710.23639-8-david@redhat.com>
Signed-off-by: NCornelia Huck <cohuck@redhat.com>

We can directly work on the uint64_t value, no need for a temporary
uint32_t value.

Also cleanup and shorten the comments.
Reviewed-by: NRichard Henderson <richard.henderson@linaro.org>
Signed-off-by: NDavid Hildenbrand <david@redhat.com>
Message-Id: <20190218122710.23639-7-david@redhat.com>
Signed-off-by: NCornelia Huck <cohuck@redhat.com>

IEEE underflows are not reported when the mask bit is off and we don't
also have an inexact exception.

z14 PoP, 9-20, "IEEE Underflow":
    An IEEE-underflow exception is recognized for an
    IEEE target when the tininess condition exists and
    either: (1) the IEEE-underflow mask bit in the FPC
    register is zero and the result value is inexact, or (2)
    the IEEE-underflow mask bit in the FPC register is
    one.
Reviewed-by: NRichard Henderson <richard.henderson@linaro.org>
Signed-off-by: NDavid Hildenbrand <david@redhat.com>
Message-Id: <20190218122710.23639-6-david@redhat.com>
Signed-off-by: NCornelia Huck <cohuck@redhat.com>

Many things are wrong and some parts cannot be fixed yet. Fix what we
can fix easily and add two FIXMEs:

The fpc flags are not updated in case an exception is actually injected.
Inexact exceptions have to be handled separately, as they are the only
exceptions that can coexist with underflows and overflows.

I reread the horribly complicated chapters in the PoP at least 5 times
and hope I got it right.

For references:
- z14 PoP, 9-18, "IEEE Exceptions"
- z14 PoP, 19-9, Figure 19-8
Reviewed-by: NRichard Henderson <richard.henderson@linaro.org>
Signed-off-by: NDavid Hildenbrand <david@redhat.com>
Message-Id: <20190218122710.23639-5-david@redhat.com>
Signed-off-by: NCornelia Huck <cohuck@redhat.com>

We want to reuse that function in vector instruction context. While at it,
cleanup the code, using defines for magic values and avoiding the
handcrafted bit conversion.
Reviewed-by: NRichard Henderson <richard.henderson@linaro.org>
Signed-off-by: NDavid Hildenbrand <david@redhat.com>
Message-Id: <20190218122710.23639-4-david@redhat.com>
Signed-off-by: NCornelia Huck <cohuck@redhat.com>

Let's use the proper conversion functions now that we have them.
Reviewed-by: NRichard Henderson <richard.henderson@linaro.org>
Signed-off-by: NDavid Hildenbrand <david@redhat.com>
Message-Id: <20190218122710.23639-3-david@redhat.com>
Signed-off-by: NCornelia Huck <cohuck@redhat.com>

Let's detect normal and denormal ("subnormal") numbers reliably. Also
test for quiet NaN's. As only one class is possible, test common cases
first.

While at it, use a better check to test for the mask bits in the data
class mask. The data class mask has 12 bits, whereby bit 0 is the
leftmost bit and bit 11 the rightmost bit. In the PoP an easy to read
table with the numbers is provided for the VECTOR FP TEST DATA CLASS
IMMEDIATE instruction, the table for TEST DATA CLASS is more confusing
as it is based on 64 bit values.

Factor the checks out into separate functions, as they will also be
needed for floating point vector instructions. We can use a makro to
generate the functions.
Signed-off-by: NDavid Hildenbrand <david@redhat.com>
Message-Id: <20190218122710.23639-2-david@redhat.com>
Reviewed-by: NRichard Henderson <richard.henderson@linaro.org>
Signed-off-by: NCornelia Huck <cohuck@redhat.com>

Use a new CC helper to calculate the CC lazily if needed. While the
PoP mentions that "A 32-bit unsigned binary integer" is placed into the
first operand, there is no word telling that the other 32 bits (high
part) are left untouched. Maybe the other 32-bit are unpredictable.
So store 64 bit for now.

Bit magic courtesy of Richard.
Signed-off-by: NDavid Hildenbrand <david@redhat.com>
Message-Id: <20190225200318.16102-8-david@redhat.com>
Reviewed-by: NRichard Henderson <richard.henderson@linaro.org>
Signed-off-by: NCornelia Huck <cohuck@redhat.com>