gcc-5 compile error in 62.0.3202.18

[Paweł Hajdan, Jr.]

M62 introduced some change that I'm not sure how to make compile with gcc-5. Does anyone have ideas how to make gcc-5.4.0 compile this code?

In file included from ../../gpu/ipc/common/mailbox_holder_struct_traits.h:10:0,

                 from ../../cc/ipc/texture_mailbox_struct_traits.h:10,

                 from ../../cc/ipc/copy_output_result_struct_traits.h:10,

                 from ../../cc/ipc/copy_output_result_struct_traits.cc:5:

../../gpu/ipc/common/mailbox_struct_traits.h: In static member function ‘static base::span<const signed char> mojo::StructTraits<gpu::mojom::MailboxDataView, gpu::Mailbox>::name(const gpu::Mailbox&)’:

../../gpu/ipc/common/mailbox_struct_traits.h:18:20: error: could not convert ‘(const int8_t*)(& mailbox.gpu::Mailbox::name)’ from ‘const int8_t* {aka const signed char*}’ to ‘base::span<const signed char>’

     return mailbox.name;

                    ^

Paweł

[Ken Rockot]

Looks like an issue with base::span template magic. +dcheng who I believe changed this and introduced base::span.

The gist is that mailbox.name is an int8_t[] and gcc is probably failing to resolve the correct span constructor (the one templated over a constant array size). A workaround should be:

  return base::span<const int8_t>(&mailbox.name[0], GL_MAILBOX_SIZE_CHROMIUM);

--
--
Chromium Developers mailing list: chromi...@chromium.org
View archives, change email options, or unsubscribe:
http://groups.google.com/a/chromium.org/group/chromium-dev
---
You received this message because you are subscribed to the Google Groups "Chromium-dev" group.
To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/chromium-dev/CAATLsPbmfm9vbn3tmCvvYzkG1SF3N1XgXeWMS_ai1DPkPye%2Bug%40mail.gmail.com.

[Paweł Hajdan, Jr.]

Ah, thanks!

I came up with the following diff. Does it look OK? I probably wouldn't upstream that, but distros using older gcc could carry this as local patch. I'd still want to confirm its sanity and correctness with you.

--- a/gpu/ipc/common/mailbox_struct_traits.h

+++ b/gpu/ipc/common/mailbox_struct_traits.h

@@ -15,7 +15,7 @@ namespace mojo {

 template <>

 struct StructTraits<gpu::mojom::MailboxDataView, gpu::Mailbox> {

   static base::span<const int8_t> name(const gpu::Mailbox& mailbox) {

-    return mailbox.name;

+    return base::span<const int8_t>(&mailbox.name[0], GL_MAILBOX_SIZE_CHROMIUM);

   }

   static bool Read(gpu::mojom::MailboxDataView data, gpu::Mailbox* out);

 };

--- a/services/viz/public/cpp/compositing/filter_operation_struct_traits.h

+++ b/services/viz/public/cpp/compositing/filter_operation_struct_traits.h

@@ -134,7 +134,7 @@ struct StructTraits<viz::mojom::FilterOperationDataView, cc::FilterOperation> {

   static base::span<const float> matrix(const cc::FilterOperation& operation) {

     if (operation.type() != cc::FilterOperation::COLOR_MATRIX)

       return base::span<const float>();

-    return operation.matrix();

+    return base::span<const float>(operation.matrix());

   }

   static base::span<const gfx::Rect> shape(

--- a/services/viz/public/cpp/compositing/quads_struct_traits.h

+++ b/services/viz/public/cpp/compositing/quads_struct_traits.h

@@ -303,7 +303,7 @@ struct StructTraits<viz::mojom::TextureQuadStateDataView, viz::DrawQuad> {

   static base::span<const float> vertex_opacity(const viz::DrawQuad& input) {

     const viz::TextureDrawQuad* quad =

         viz::TextureDrawQuad::MaterialCast(&input);

-    return quad->vertex_opacity;

+    return base::span<const float>(quad->vertex_opacity);

   }

   static bool y_flipped(const viz::DrawQuad& input) {

--- a/third_party/WebKit/Source/platform/exported/WebCORS.cpp

+++ b/third_party/WebKit/Source/platform/exported/WebCORS.cpp

@@ -480,7 +480,7 @@ WebString AccessControlErrorString(

     }

     default:

       NOTREACHED();

-      return "";

+      return WebString();

   }

 }

@@ -512,7 +512,7 @@ WebString PreflightErrorString(const PreflightStatus status,

     }

     default:

       NOTREACHED();

-      return "";

+      return WebString();

   }

 }

@@ -533,7 +533,7 @@ WebString RedirectErrorString(const RedirectStatus status,

     }

     default:

       NOTREACHED();

-      return "";

+      return WebString();

   }

 }

Paweł

[Ken Rockot]

Looks OK but given the other two base::span changes it kind of looks explicit construction should also work for Mailbox, i.e.:

  return base::span<const int8_t>(mailbox.name);

Is that not the case?

[Marijn Kruisselbrink]

On Thu, Sep 28, 2017 at 12:04 PM, Ken Rockot <roc...@chromium.org> wrote:

Looks OK but given the other two base::span changes it kind of looks explicit construction should also work for Mailbox, i.e.:

  return base::span<const int8_t>(mailbox.name);

Or maybe even base::make_span(mailbox.name), avoiding the explicit type entirely?

 

To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/chromium-dev/CA%2BapAgGZP2yO1K5m%3DkBpJhCsWkMpXP%2B4rCJT1X9xUUUODPs0SA%40mail.gmail.com.

[Paweł Hajdan, Jr.]

Yup, that works.

Thanks!

Paweł

[Daniel Cheng]

Oops, sorry for missing this thread--I was travelling. Pawel, changing it to use base::make_span looks fine. I think it's a defect in gcc5 and how it looks up conversion constructors, but I wasn't able to find a bug number (and it seems to be fixed for gcc6 anyway).

Daniel