[Buildroot] [PATCH 1/2] toolchain: introduce BR2_TOOLCHAIN_HAS_BINUTILS_BUG_21464

[Buildroot] [PATCH 1/2] toolchain: introduce BR2_TOOLCHAIN_HAS_BINUTILS_BUG_21464

[Giulio Benetti]

On OpenRisc binutils it still present ld bug 21464 leading to a package
protobuf to fail building:
http://autobuild.buildroot.net/results/908/9084cd777aefe0fa8235514c33767d8640ad7a5b/

The bug was already reported and it's been updated:
https://sourceware.org/bugzilla/show_bug.cgi?id=21464

Signed-off-by: Giulio Benetti <giulio.benetti@benettiengineering.com>
---
 toolchain/Config.in | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/toolchain/Config.in b/toolchain/Config.in
index 87509f3d64..d71bb2a65a 100644
--- a/toolchain/Config.in
+++ b/toolchain/Config.in
@@ -81,6 +81,12 @@ config BR2_TOOLCHAIN_HAS_BINUTILS_BUG_19615
 config BR2_TOOLCHAIN_HAS_BINUTILS_BUG_20006
 	bool
 
+# https://sourceware.org/bugzilla/show_bug.cgi?id=21464
+# Affect toolchains built with binutils 2.31.1, still not fixed.
+config BR2_TOOLCHAIN_HAS_BINUTILS_BUG_21464
+	bool
+	default y if BR2_or1k
+
 # Atomic types can be:
 #  - never lock-free
 #  - sometimes lock-free
-- 
2.20.1

[Buildroot] [PATCH 2/2] package/protobuf: disable package if binutils is affected from bug 21464

[Giulio Benetti]

This package is affected from binutils bug 21464, since there is no
workaround, let's disable it.

Fixes:
http://autobuild.buildroot.net/results/908/9084cd777aefe0fa8235514c33767d8640ad7a5b/

Signed-off-by: Giulio Benetti <giulio.benetti@benettiengineering.com>
---
 package/protobuf/Config.in | 1 +
 1 file changed, 1 insertion(+)

diff --git a/package/protobuf/Config.in b/package/protobuf/Config.in
index 46c820b9cf..30050950aa 100644
--- a/package/protobuf/Config.in
+++ b/package/protobuf/Config.in
@@ -24,6 +24,7 @@ config BR2_PACKAGE_PROTOBUF_ARCH_SUPPORTS
 	default y if BR2_TOOLCHAIN_HAS_ATOMIC
 	depends on BR2_HOSTARCH = "x86_64" || BR2_HOSTARCH = "x86"
 	depends on BR2_USE_MMU # fork()
+	depends on !BR2_TOOLCHAIN_HAS_BINUTILS_BUG_21464
 
 config BR2_PACKAGE_PROTOBUF
 	bool "protobuf"
-- 
2.20.1

[Buildroot] [PATCH 1/2] toolchain: introduce BR2_TOOLCHAIN_HAS_BINUTILS_BUG_21464

[Thomas Petazzoni]

On Fri, 28 Feb 2020 18:58:13 +0100
Giulio Benetti <giulio.benetti@benettiengineering.com> wrote:

> On OpenRisc binutils it still present ld bug 21464 leading to a package
> protobuf to fail building:
> http://autobuild.buildroot.net/results/908/9084cd777aefe0fa8235514c33767d8640ad7a5b/
> 
> The bug was already reported and it's been updated:
> https://sourceware.org/bugzilla/show_bug.cgi?id=21464
> 
> Signed-off-by: Giulio Benetti <giulio.benetti@benettiengineering.com>
> ---
>  toolchain/Config.in | 6 ++++++
>  1 file changed, 6 insertions(+)
> 
> diff --git a/toolchain/Config.in b/toolchain/Config.in
> index 87509f3d64..d71bb2a65a 100644
> --- a/toolchain/Config.in
> +++ b/toolchain/Config.in
> @@ -81,6 +81,12 @@ config BR2_TOOLCHAIN_HAS_BINUTILS_BUG_19615
>  config BR2_TOOLCHAIN_HAS_BINUTILS_BUG_20006
>  	bool
>  
> +# https://sourceware.org/bugzilla/show_bug.cgi?id=21464
> +# Affect toolchains built with binutils 2.31.1, still not fixed.
> +config BR2_TOOLCHAIN_HAS_BINUTILS_BUG_21464
> +	bool
> +	default y if BR2_or1k

The problem here is that we don't have any Config.in options for
binutils versions, so should this problem be fixed one day, we would
have no way to update this Config.in option.

To be honest, I am starting to wonder if we should keep support for
OpenRISC. It's using a now quite old gcc version, and I'm not sure the
OpenRISC support has been upstreamed to gcc.

Also, we don't have anyone specifically interested in OpenRISC and
maintaining it...

Thomas
-- 
Thomas Petazzoni, CTO, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com

[Buildroot] [PATCH 1/2] toolchain: introduce BR2_TOOLCHAIN_HAS_BINUTILS_BUG_21464

[Giulio Benetti]

Hi Thomas, All,

On 2/28/20 11:19 PM, Thomas Petazzoni wrote:
> On Fri, 28 Feb 2020 18:58:13 +0100
> Giulio Benetti <giulio.benetti@benettiengineering.com> wrote:
> 
>> On OpenRisc binutils it still present ld bug 21464 leading to a package
>> protobuf to fail building:
>> http://autobuild.buildroot.net/results/908/9084cd777aefe0fa8235514c33767d8640ad7a5b/
>>
>> The bug was already reported and it's been updated:
>> https://sourceware.org/bugzilla/show_bug.cgi?id=21464
>>
>> Signed-off-by: Giulio Benetti <giulio.benetti@benettiengineering.com>
>> ---
>>   toolchain/Config.in | 6 ++++++
>>   1 file changed, 6 insertions(+)
>>
>> diff --git a/toolchain/Config.in b/toolchain/Config.in
>> index 87509f3d64..d71bb2a65a 100644
>> --- a/toolchain/Config.in
>> +++ b/toolchain/Config.in
>> @@ -81,6 +81,12 @@ config BR2_TOOLCHAIN_HAS_BINUTILS_BUG_19615
>>   config BR2_TOOLCHAIN_HAS_BINUTILS_BUG_20006
>>   	bool
>>   
>> +# https://sourceware.org/bugzilla/show_bug.cgi?id=21464
>> +# Affect toolchains built with binutils 2.31.1, still not fixed.
>> +config BR2_TOOLCHAIN_HAS_BINUTILS_BUG_21464
>> +	bool
>> +	default y if BR2_or1k
> 
> The problem here is that we don't have any Config.in options for
> binutils versions, so should this problem be fixed one day, we would
> have no way to update this Config.in option.

So at the moment it would make sense to only specify in protobuf 
Config.in that or1k is not supported, right?

> To be honest, I am starting to wonder if we should keep support for
> OpenRISC. It's using a now quite old gcc version, and I'm not sure the
> OpenRISC support has been upstreamed to gcc.
> 
> Also, we don't have anyone specifically interested in OpenRISC and
> maintaining it...

I don't know, is it more used than Microblaze or less? Maybe that could 
give the idea if keep it or not after all the work with gcc workarounds 
for Microblaze IMHO.

-- 
Giulio Benetti
Benetti Engineering sas