FAQ
I'm still hung up on Windows, but I tried running ssa with the
warn-about-unimplemented flag enabled,
and this is what apparently remains from running all.bash,
with an early out in the test subdirectory because the chattiness failed
some of the error-output checks:

grep "SSA unimpl" all.log | sort | uniq -c | sort -nr

2832 SSA unimplemented: questionable CONVNOP from race detector
unsafe.Pointer -> uintptr
2496 SSA unimplemented: unhandled closure arg
  612 SSA unimplemented: race detector mishandles nodfp
  554 SSA unimplemented: not lowered: Load STRUCT PTR64 mem
   60 SSA unimplemented: questionable CONVNOP from race detector int -> int64
   27 SSA unimplemented: local variable with class PPARAMOUT,heap
unimplemented
   10 SSA unimplemented: questionable CONVNOP from race detector
reflect.flag -> reflect.Kind
   10 SSA unimplemented: not lowered: Store mem PTR64 STRUCT mem
    6 SSA unimplemented: not lowered: Load ARRAY PTR64 mem
    2 SSA unimplemented: questionable CONVNOP from race detector uint64 ->
Word
    2 SSA unimplemented: questionable CONVNOP from race detector
map[string]string -> *uint8
    2 SSA unimplemented: questionable CONVNOP from race detector
map[string]*Flag -> *uint8
    2 SSA unimplemented: questionable CONVNOP from race detector _Ctype_long
-> int
    2 SSA unimplemented: not lowered: Store mem PTR64 ARRAY mem
    1 SSA unimplemented: questionable CONVNOP from race detector uint ->
uint64
    1 SSA unimplemented: questionable CONVNOP from race detector
_Ctype_size_t -> uintptr
    1 SSA unimplemented: questionable CONVNOP from race detector Header ->
*uint8

I'm going to set the Windows/runtime problem aside for now, and look into
2496 SSA unimplemented: unhandled closure arg
   27 SSA unimplemented: local variable with class PPARAMOUT,heap
unimplemented

--
You received this message because you are subscribed to the Google Groups "golang-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to golang-dev+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Search Discussions

  • Keith Randall at Oct 5, 2015 at 4:52 pm
    I'm pretty sure I know what is causing the "not lowered: Load/Store"
    failures. These are loads/stores of large objects (structs and arrays)
    where the load and store can't be matched up to generate a mem->mem move.
    I'm investigating.

    On Mon, Oct 5, 2015 at 9:38 AM, 'David Chase' via golang-dev wrote:

    I'm still hung up on Windows, but I tried running ssa with the
    warn-about-unimplemented flag enabled,
    and this is what apparently remains from running all.bash,
    with an early out in the test subdirectory because the chattiness failed
    some of the error-output checks:

    grep "SSA unimpl" all.log | sort | uniq -c | sort -nr

    2832 SSA unimplemented: questionable CONVNOP from race detector
    unsafe.Pointer -> uintptr
    2496 SSA unimplemented: unhandled closure arg
    612 SSA unimplemented: race detector mishandles nodfp
    554 SSA unimplemented: not lowered: Load STRUCT PTR64 mem
    60 SSA unimplemented: questionable CONVNOP from race detector int ->
    int64
    27 SSA unimplemented: local variable with class PPARAMOUT,heap
    unimplemented
    10 SSA unimplemented: questionable CONVNOP from race detector
    reflect.flag -> reflect.Kind
    10 SSA unimplemented: not lowered: Store mem PTR64 STRUCT mem
    6 SSA unimplemented: not lowered: Load ARRAY PTR64 mem
    2 SSA unimplemented: questionable CONVNOP from race detector uint64 ->
    Word
    2 SSA unimplemented: questionable CONVNOP from race detector
    map[string]string -> *uint8
    2 SSA unimplemented: questionable CONVNOP from race detector
    map[string]*Flag -> *uint8
    2 SSA unimplemented: questionable CONVNOP from race detector
    _Ctype_long -> int
    2 SSA unimplemented: not lowered: Store mem PTR64 ARRAY mem
    1 SSA unimplemented: questionable CONVNOP from race detector uint ->
    uint64
    1 SSA unimplemented: questionable CONVNOP from race detector
    _Ctype_size_t -> uintptr
    1 SSA unimplemented: questionable CONVNOP from race detector Header ->
    *uint8

    I'm going to set the Windows/runtime problem aside for now, and look into
    2496 SSA unimplemented: unhandled closure arg
    27 SSA unimplemented: local variable with class PPARAMOUT,heap
    unimplemented

    --
    You received this message because you are subscribed to the Google Groups
    "golang-dev" group.
    To unsubscribe from this group and stop receiving emails from it, send an
    email to golang-dev+unsubscribe@googlegroups.com.
    For more options, visit https://groups.google.com/d/optout.
    --
    You received this message because you are subscribed to the Google Groups "golang-dev" group.
    To unsubscribe from this group and stop receiving emails from it, send an email to golang-dev+unsubscribe@googlegroups.com.
    For more options, visit https://groups.google.com/d/optout.
  • David Chase at Oct 5, 2015 at 7:53 pm
    Found a minor bit of WTF here. For one of the "SSA unimplemented: local
    variable with class PPARAMOUT,heap unimplemented" bugs, when I clear that
    case, I later fail here. If you look closely at the input, you'll see that
    "val" is initialized to zero twice (init-to-zero = single-operand-AS), once
    with class PPARAM, once with class PPARAMOUT. Is this a bug or a feature
    of the input tree? (i.e., shouldn't the input say PPARAMOUT both times?)

    /Users/drchase/GoogleDrive/work/go-ssa/src/reflect/value.go:1287: internal
    compiler error: addr of undeclared ONAME val. declared: map[v:v4 nb:v5
    .fp:v6]
    buildssa-enter
    . AS l(3731)
    . . NAME-reflect.val a(true) l(3731) x(32+0) class(PPARAM) tc(1)
    reflect.Value

    . AS l(3731)
    . . NAME-reflect.&val u(1) a(true) l(4974) x(0+0) class(PAUTO) esc(N)
    tc(1) used(true) PTR64-*Value
    . . CALLFUNC u(100) l(3731) tc(1) PTR64-*Value
    . . . NAME-runtime.newobject u(1) a(true) l(2) x(0+0) class(PFUNC)
    tc(1) used(true) FUNC-func(*byte) *Value
    . . CALLFUNC-list
    . . . AS u(2) l(3731) tc(1)
    . . . . INDREG-SP a(true) l(3731) x(0+0) tc(1) runtime.typ·2
    PTR64-*byte
    . . . . ADDR u(2) a(true) l(3731) tc(1) PTR64-*uint8
    . . . . . NAME-type."".Value u(1) a(true) l(24) x(0+0)
    class(PEXTERN) tc(1) used(true) uint8

    . AS l(3731)
    . . NAME-reflect.ok u(1) a(true) g(2) l(3731) x(56+0) class(PPARAMOUT)
    f(1) tc(1) assigned bool
    buildssa-body
    . AS u(2) l(3731) tc(1)
    . . NAME-reflect.ok u(1) a(true) g(2) l(3731) x(56+0) class(PPARAMOUT)
    f(1) tc(1) assigned bool

    . AS u(2) l(3731) tc(1)
    . . NAME-reflect.val u(2) g(1) l(3731) x(0+0) class(PPARAMOUT,heap)
    f(1) esc(h) tc(1) addrtaken assigned used(true) reflect.Value

    . DCL l(3732)
    . . NAME-reflect.tt u(1) a(true) g(5) l(3732) x(0+0) class(PAUTO) f(1)
    tc(1) used(true) PTR64-*chanType

    On Mon, Oct 5, 2015 at 12:52 PM, Keith Randall wrote:

    I'm pretty sure I know what is causing the "not lowered: Load/Store"
    failures. These are loads/stores of large objects (structs and arrays)
    where the load and store can't be matched up to generate a mem->mem move.
    I'm investigating.


    On Mon, Oct 5, 2015 at 9:38 AM, 'David Chase' via golang-dev <
    golang-dev@googlegroups.com> wrote:
    I'm still hung up on Windows, but I tried running ssa with the
    warn-about-unimplemented flag enabled,
    and this is what apparently remains from running all.bash,
    with an early out in the test subdirectory because the chattiness failed
    some of the error-output checks:

    grep "SSA unimpl" all.log | sort | uniq -c | sort -nr

    2832 SSA unimplemented: questionable CONVNOP from race detector
    unsafe.Pointer -> uintptr
    2496 SSA unimplemented: unhandled closure arg
    612 SSA unimplemented: race detector mishandles nodfp
    554 SSA unimplemented: not lowered: Load STRUCT PTR64 mem
    60 SSA unimplemented: questionable CONVNOP from race detector int ->
    int64
    27 SSA unimplemented: local variable with class PPARAMOUT,heap
    unimplemented
    10 SSA unimplemented: questionable CONVNOP from race detector
    reflect.flag -> reflect.Kind
    10 SSA unimplemented: not lowered: Store mem PTR64 STRUCT mem
    6 SSA unimplemented: not lowered: Load ARRAY PTR64 mem
    2 SSA unimplemented: questionable CONVNOP from race detector uint64 ->
    Word
    2 SSA unimplemented: questionable CONVNOP from race detector
    map[string]string -> *uint8
    2 SSA unimplemented: questionable CONVNOP from race detector
    map[string]*Flag -> *uint8
    2 SSA unimplemented: questionable CONVNOP from race detector
    _Ctype_long -> int
    2 SSA unimplemented: not lowered: Store mem PTR64 ARRAY mem
    1 SSA unimplemented: questionable CONVNOP from race detector uint ->
    uint64
    1 SSA unimplemented: questionable CONVNOP from race detector
    _Ctype_size_t -> uintptr
    1 SSA unimplemented: questionable CONVNOP from race detector Header ->
    *uint8

    I'm going to set the Windows/runtime problem aside for now, and look into
    2496 SSA unimplemented: unhandled closure arg
    27 SSA unimplemented: local variable with class PPARAMOUT,heap
    unimplemented

    --
    You received this message because you are subscribed to the Google Groups
    "golang-dev" group.
    To unsubscribe from this group and stop receiving emails from it, send an
    email to golang-dev+unsubscribe@googlegroups.com.
    For more options, visit https://groups.google.com/d/optout.
    --
    You received this message because you are subscribed to the Google Groups "golang-dev" group.
    To unsubscribe from this group and stop receiving emails from it, send an email to golang-dev+unsubscribe@googlegroups.com.
    For more options, visit https://groups.google.com/d/optout.
  • Keith Randall at Oct 5, 2015 at 8:14 pm
    I don't understand why it is not PPARAMOUT in both places either.

    On Mon, Oct 5, 2015 at 12:53 PM, David Chase wrote:

    Found a minor bit of WTF here. For one of the "SSA unimplemented: local
    variable with class PPARAMOUT,heap unimplemented" bugs, when I clear that
    case, I later fail here. If you look closely at the input, you'll see that
    "val" is initialized to zero twice (init-to-zero = single-operand-AS), once
    with class PPARAM, once with class PPARAMOUT. Is this a bug or a feature
    of the input tree? (i.e., shouldn't the input say PPARAMOUT both times?)

    /Users/drchase/GoogleDrive/work/go-ssa/src/reflect/value.go:1287: internal
    compiler error: addr of undeclared ONAME val. declared: map[v:v4 nb:v5
    .fp:v6]
    buildssa-enter
    . AS l(3731)
    . . NAME-reflect.val a(true) l(3731) x(32+0) class(PPARAM) tc(1)
    reflect.Value

    . AS l(3731)
    . . NAME-reflect.&val u(1) a(true) l(4974) x(0+0) class(PAUTO) esc(N)
    tc(1) used(true) PTR64-*Value
    . . CALLFUNC u(100) l(3731) tc(1) PTR64-*Value
    . . . NAME-runtime.newobject u(1) a(true) l(2) x(0+0) class(PFUNC)
    tc(1) used(true) FUNC-func(*byte) *Value
    . . CALLFUNC-list
    . . . AS u(2) l(3731) tc(1)
    . . . . INDREG-SP a(true) l(3731) x(0+0) tc(1) runtime.typ·2
    PTR64-*byte
    . . . . ADDR u(2) a(true) l(3731) tc(1) PTR64-*uint8
    . . . . . NAME-type."".Value u(1) a(true) l(24) x(0+0)
    class(PEXTERN) tc(1) used(true) uint8

    . AS l(3731)
    . . NAME-reflect.ok u(1) a(true) g(2) l(3731) x(56+0) class(PPARAMOUT)
    f(1) tc(1) assigned bool
    buildssa-body
    . AS u(2) l(3731) tc(1)
    . . NAME-reflect.ok u(1) a(true) g(2) l(3731) x(56+0) class(PPARAMOUT)
    f(1) tc(1) assigned bool

    . AS u(2) l(3731) tc(1)
    . . NAME-reflect.val u(2) g(1) l(3731) x(0+0) class(PPARAMOUT,heap)
    f(1) esc(h) tc(1) addrtaken assigned used(true) reflect.Value

    . DCL l(3732)
    . . NAME-reflect.tt u(1) a(true) g(5) l(3732) x(0+0) class(PAUTO) f(1)
    tc(1) used(true) PTR64-*chanType

    On Mon, Oct 5, 2015 at 12:52 PM, Keith Randall wrote:

    I'm pretty sure I know what is causing the "not lowered: Load/Store"
    failures. These are loads/stores of large objects (structs and arrays)
    where the load and store can't be matched up to generate a mem->mem move.
    I'm investigating.


    On Mon, Oct 5, 2015 at 9:38 AM, 'David Chase' via golang-dev <
    golang-dev@googlegroups.com> wrote:
    I'm still hung up on Windows, but I tried running ssa with the
    warn-about-unimplemented flag enabled,
    and this is what apparently remains from running all.bash,
    with an early out in the test subdirectory because the chattiness failed
    some of the error-output checks:

    grep "SSA unimpl" all.log | sort | uniq -c | sort -nr

    2832 SSA unimplemented: questionable CONVNOP from race detector
    unsafe.Pointer -> uintptr
    2496 SSA unimplemented: unhandled closure arg
    612 SSA unimplemented: race detector mishandles nodfp
    554 SSA unimplemented: not lowered: Load STRUCT PTR64 mem
    60 SSA unimplemented: questionable CONVNOP from race detector int ->
    int64
    27 SSA unimplemented: local variable with class PPARAMOUT,heap
    unimplemented
    10 SSA unimplemented: questionable CONVNOP from race detector
    reflect.flag -> reflect.Kind
    10 SSA unimplemented: not lowered: Store mem PTR64 STRUCT mem
    6 SSA unimplemented: not lowered: Load ARRAY PTR64 mem
    2 SSA unimplemented: questionable CONVNOP from race detector uint64
    -> Word
    2 SSA unimplemented: questionable CONVNOP from race detector
    map[string]string -> *uint8
    2 SSA unimplemented: questionable CONVNOP from race detector
    map[string]*Flag -> *uint8
    2 SSA unimplemented: questionable CONVNOP from race detector
    _Ctype_long -> int
    2 SSA unimplemented: not lowered: Store mem PTR64 ARRAY mem
    1 SSA unimplemented: questionable CONVNOP from race detector uint ->
    uint64
    1 SSA unimplemented: questionable CONVNOP from race detector
    _Ctype_size_t -> uintptr
    1 SSA unimplemented: questionable CONVNOP from race detector Header
    -> *uint8

    I'm going to set the Windows/runtime problem aside for now, and look
    into
    2496 SSA unimplemented: unhandled closure arg
    27 SSA unimplemented: local variable with class PPARAMOUT,heap
    unimplemented

    --
    You received this message because you are subscribed to the Google
    Groups "golang-dev" group.
    To unsubscribe from this group and stop receiving emails from it, send
    an email to golang-dev+unsubscribe@googlegroups.com.
    For more options, visit https://groups.google.com/d/optout.
    --
    You received this message because you are subscribed to the Google Groups "golang-dev" group.
    To unsubscribe from this group and stop receiving emails from it, send an email to golang-dev+unsubscribe@googlegroups.com.
    For more options, visit https://groups.google.com/d/optout.
  • David Chase at Oct 7, 2015 at 9:16 pm
    Progress on that (it's a glitch in the old compiler, fixable), and in the
    process of figuring out the next downstream bug, discovered that we're not
    placing our write barriers correctly.
    On Mon, Oct 5, 2015 at 4:14 PM, Keith Randall wrote:

    I don't understand why it is not PPARAMOUT in both places either.

    On Mon, Oct 5, 2015 at 12:53 PM, David Chase wrote:

    Found a minor bit of WTF here. For one of the "SSA unimplemented: local
    variable with class PPARAMOUT,heap unimplemented" bugs, when I clear that
    case, I later fail here. If you look closely at the input, you'll see that
    "val" is initialized to zero twice (init-to-zero = single-operand-AS), once
    with class PPARAM, once with class PPARAMOUT. Is this a bug or a feature
    of the input tree? (i.e., shouldn't the input say PPARAMOUT both times?)

    /Users/drchase/GoogleDrive/work/go-ssa/src/reflect/value.go:1287:
    internal compiler error: addr of undeclared ONAME val. declared: map[v:v4
    nb:v5 .fp:v6]
    buildssa-enter
    . AS l(3731)
    . . NAME-reflect.val a(true) l(3731) x(32+0) class(PPARAM) tc(1)
    reflect.Value

    . AS l(3731)
    . . NAME-reflect.&val u(1) a(true) l(4974) x(0+0) class(PAUTO) esc(N)
    tc(1) used(true) PTR64-*Value
    . . CALLFUNC u(100) l(3731) tc(1) PTR64-*Value
    . . . NAME-runtime.newobject u(1) a(true) l(2) x(0+0) class(PFUNC)
    tc(1) used(true) FUNC-func(*byte) *Value
    . . CALLFUNC-list
    . . . AS u(2) l(3731) tc(1)
    . . . . INDREG-SP a(true) l(3731) x(0+0) tc(1) runtime.typ·2
    PTR64-*byte
    . . . . ADDR u(2) a(true) l(3731) tc(1) PTR64-*uint8
    . . . . . NAME-type."".Value u(1) a(true) l(24) x(0+0)
    class(PEXTERN) tc(1) used(true) uint8

    . AS l(3731)
    . . NAME-reflect.ok u(1) a(true) g(2) l(3731) x(56+0)
    class(PPARAMOUT) f(1) tc(1) assigned bool
    buildssa-body
    . AS u(2) l(3731) tc(1)
    . . NAME-reflect.ok u(1) a(true) g(2) l(3731) x(56+0)
    class(PPARAMOUT) f(1) tc(1) assigned bool

    . AS u(2) l(3731) tc(1)
    . . NAME-reflect.val u(2) g(1) l(3731) x(0+0) class(PPARAMOUT,heap)
    f(1) esc(h) tc(1) addrtaken assigned used(true) reflect.Value

    . DCL l(3732)
    . . NAME-reflect.tt u(1) a(true) g(5) l(3732) x(0+0) class(PAUTO)
    f(1) tc(1) used(true) PTR64-*chanType

    On Mon, Oct 5, 2015 at 12:52 PM, Keith Randall wrote:

    I'm pretty sure I know what is causing the "not lowered: Load/Store"
    failures. These are loads/stores of large objects (structs and arrays)
    where the load and store can't be matched up to generate a mem->mem move.
    I'm investigating.


    On Mon, Oct 5, 2015 at 9:38 AM, 'David Chase' via golang-dev <
    golang-dev@googlegroups.com> wrote:
    I'm still hung up on Windows, but I tried running ssa with the
    warn-about-unimplemented flag enabled,
    and this is what apparently remains from running all.bash,
    with an early out in the test subdirectory because the chattiness
    failed some of the error-output checks:

    grep "SSA unimpl" all.log | sort | uniq -c | sort -nr

    2832 SSA unimplemented: questionable CONVNOP from race detector
    unsafe.Pointer -> uintptr
    2496 SSA unimplemented: unhandled closure arg
    612 SSA unimplemented: race detector mishandles nodfp
    554 SSA unimplemented: not lowered: Load STRUCT PTR64 mem
    60 SSA unimplemented: questionable CONVNOP from race detector int ->
    int64
    27 SSA unimplemented: local variable with class PPARAMOUT,heap
    unimplemented
    10 SSA unimplemented: questionable CONVNOP from race detector
    reflect.flag -> reflect.Kind
    10 SSA unimplemented: not lowered: Store mem PTR64 STRUCT mem
    6 SSA unimplemented: not lowered: Load ARRAY PTR64 mem
    2 SSA unimplemented: questionable CONVNOP from race detector uint64
    -> Word
    2 SSA unimplemented: questionable CONVNOP from race detector
    map[string]string -> *uint8
    2 SSA unimplemented: questionable CONVNOP from race detector
    map[string]*Flag -> *uint8
    2 SSA unimplemented: questionable CONVNOP from race detector
    _Ctype_long -> int
    2 SSA unimplemented: not lowered: Store mem PTR64 ARRAY mem
    1 SSA unimplemented: questionable CONVNOP from race detector uint ->
    uint64
    1 SSA unimplemented: questionable CONVNOP from race detector
    _Ctype_size_t -> uintptr
    1 SSA unimplemented: questionable CONVNOP from race detector Header
    -> *uint8

    I'm going to set the Windows/runtime problem aside for now, and look
    into
    2496 SSA unimplemented: unhandled closure arg
    27 SSA unimplemented: local variable with class PPARAMOUT,heap
    unimplemented

    --
    You received this message because you are subscribed to the Google
    Groups "golang-dev" group.
    To unsubscribe from this group and stop receiving emails from it, send
    an email to golang-dev+unsubscribe@googlegroups.com.
    For more options, visit https://groups.google.com/d/optout.
    --
    You received this message because you are subscribed to the Google Groups "golang-dev" group.
    To unsubscribe from this group and stop receiving emails from it, send an email to golang-dev+unsubscribe@googlegroups.com.
    For more options, visit https://groups.google.com/d/optout.

Related Discussions

Discussion Navigation
viewthread | post
Discussion Overview
groupgolang-dev @
categoriesgo
postedOct 5, '15 at 4:38p
activeOct 7, '15 at 9:16p
posts5
users2
websitegolang.org

2 users in discussion

David Chase: 3 posts Keith Randall: 2 posts

People

Translate

site design / logo © 2021 Grokbase