Currently, order desugars map assignment operations like

    m[k] op= r

into

    m[k] = m[k] op r

which in turn is transformed during walk into:

    tmp := *mapaccess(m, k)
    tmp = tmp op r
    *mapassign(m, k) = tmp

However, this is suboptimal, as we could instead produce just:

    *mapassign(m, k) op= r

One complication though is if "r == 0", then "m[k] /= r" and "m[k] %=
r" will panic, and they need to do so *before* calling mapassign,
otherwise we may insert a new zero-value element into the map.

It would be spec compliant to just emit the "r != 0" check before
calling mapassign (see #23735), but currently these checks aren't
generated until SSA construction. For now, it's simpler to continue
desugaring /= and %= into two map indexing operations.

Fixes #23661.

Change-Id: I46e3739d9adef10e92b46fdd78b88d5aabe68952
Reviewed-on: https://go-review.googlesource.com/91557
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Austin Clements <austin@google.com>

In this test, traditionally the comparison `*l == r[0]` was left as a
comparison between `*l` (type `any`) and `r[0]` (type `*int`), and the
rest of the compiler needed to handle mixed-typed comparisons.
However, this means more complexity for wiring up explicit rtypes.

To simplify rtype handling, the next CL will change unified IR to
instead handle the expression as `*l == any(r[0])`. However, a
consequence of this currently is that walk will now sequence the
`any(r[0])` expression first, because it involves a
concrete-to-interface conversion. And in turn, this means the `r[0]`
panic ("index out of bounds") will take priority over the `*l`
panic ("nil pointer dereference").

This is a change in user-visible semantics in some cases, but the Go
spec leaves this unspecified, so it shouldn't be an issue. Note also:
gccgo has the same behavior (i.e., panicking on index out of bounds,
not nil pointer dereference), and cmd/compile also already has the
same behavior when the interface conversion is explicit (as in the
added "nil pointer dereference #3" test case).

Updates #23735.
Updates #32187.

Change-Id: I49e5dcca85b4680f9c8780ef0013e64254d38fe5
Reviewed-on: https://go-review.googlesource.com/c/go/+/418097
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: David Chase <drchase@google.com>

In this test, traditionally the comparison `*l == r[0]` was left as a
comparison between `*l` (type `any`) and `r[0]` (type `*int`), and the
rest of the compiler needed to handle mixed-typed comparisons.
However, this means more complexity for wiring up explicit rtypes.

To simplify rtype handling, the next CL will change unified IR to
instead handle the expression as `*l == any(r[0])`. However, a
consequence of this currently is that walk will now sequence the
`any(r[0])` expression first, because it involves a
concrete-to-interface conversion. And in turn, this means the `r[0]`
panic ("index out of bounds") will take priority over the `*l`
panic ("nil pointer dereference").

This is a change in user-visible semantics in some cases, but the Go
spec leaves this unspecified, so it shouldn't be an issue. Note also:
gccgo has the same behavior (i.e., panicking on index out of bounds,
not nil pointer dereference), and cmd/compile also already has the
same behavior when the interface conversion is explicit (as in the
added "nil pointer dereference golang#3" test case).

Updates golang#23735.
Updates golang#32187.

Change-Id: I49e5dcca85b4680f9c8780ef0013e64254d38fe5
Reviewed-on: https://go-review.googlesource.com/c/go/+/418097
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: David Chase <drchase@google.com>