Skip to contentSkip to conversation

pseudo: Fix enum typedef

PATCH In review Andreas Müller 5 participants 9 messages Jun 4, 2020

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 pseudo: Fix enum typedef

From: Jacob Kroon <[email protected]>

'pseudo_access_t' is a type, so use typedef.

Fixes building pseudo with gcc 10 where -fno-common is the default.

(Backport of OE-Core rev: a7d519f742aadc9110c2401f359254210a784f6b)

---
 .../0001-pseudo_ipc.h-Fix-enum-typedef.patch  | 31 +++++++++++++++++++
 meta/recipes-devtools/pseudo/pseudo_git.bb    |  1 +
 2 files changed, 32 insertions(+)
 create mode 100644 meta/recipes-devtools/pseudo/files/0001-pseudo_ipc.h-Fix-enum-typedef.patch
meta/recipes-devtools/pseudo/files/0001-pseudo_ipc.h-Fix-enum-typedef.patch new file +31 -0
new file mode 100644
index 00000000000..33d4ef3b2fa
@@ -0,0 +1,31 @@
+From a491aececfedf7313d29b80d626e0964fb533548 Mon Sep 17 00:00:00 2001
+From: Jacob Kroon <[email protected]>
+Date: Sun, 3 May 2020 06:24:03 +0200
+Subject: [PATCH] pseudo_ipc.h: Fix enum typedef
+
+'pseudo_access_t' is a type, so use typedef.
+
+Fixes building pseudo with gcc 10 where -fno-common is the default.
+
+Signed-off-by: Jacob Kroon <[email protected]>
+Upstream-Status: Submitted [https://lists.openembedded.org/g/openembedded-core/message/137758]
+---
+ pseudo_ipc.h | 2 +-
+ 1 file changed, 1 insertion(+), 1 deletion(-)
+
+diff --git a/pseudo_ipc.h b/pseudo_ipc.h
+index caeae5c..d945257 100644
+--- a/pseudo_ipc.h
++++ b/pseudo_ipc.h
+@@ -29,7 +29,7 @@ typedef struct {
+ 	char path[];
+ } pseudo_msg_t;
+ 
+-enum {
++typedef enum {
+ 	PSA_EXEC = 1,
+ 	PSA_WRITE = (PSA_EXEC << 1),
+ 	PSA_READ = (PSA_WRITE << 1),
+-- 
+2.26.2
+
meta/recipes-devtools/pseudo/pseudo_git.bb +1 -0
index 7db5bafd174..8ff003d9024 100644
@@ -7,6 +7,7 @@ SRC_URI = "git://git.yoctoproject.org/pseudo \
            file://moreretries.patch \
            file://toomanyfiles.patch \
            file://0001-Add-statx.patch \
+           file://0001-pseudo_ipc.h-Fix-enum-typedef.patch \
            "
 
 SRCREV = "060058bb29f70b244e685b3c704eb0641b736f73"
Signature · 1 line
2.21.3
#1 lore raw
Quoted 11 lines · Andreas Müller

From: Jacob Kroon <[email protected]>

'pseudo_access_t' is a type, so use typedef.

Fixes building pseudo with gcc 10 where -fno-common is the default.

(Backport of OE-Core rev: a7d519f742aadc9110c2401f359254210a784f6b)

Does warrior otherwise work on gcc10 hosts?

Cheers,

Richard

#2 lore raw
Quoted 8 lines · Richard Purdie
Quoted 6 lines · Andreas Müller

From: Jacob Kroon <[email protected]>

'pseudo_access_t' is a type, so use typedef.

Fixes building pseudo with gcc 10 where -fno-common is the default.

Quoted 7 lines
Quoted 5 lines

(Backport of OE-Core rev: a7d519f742aadc9110c2401f359254210a784f6b)

Does warrior otherwise work on gcc10 hosts?

Not yet but I am working on it.

Andreas

Quoted 5 lines

Cheers,

Richard

#3 lore raw
Quoted 13 lines · Andreas Müller
Quoted 11 lines · Richard Purdie
Quoted 10 lines · Andreas Müller

From: Jacob Kroon <[email protected]>

'pseudo_access_t' is a type, so use typedef.

Fixes building pseudo with gcc 10 where -fno-common is the default.

(Backport of OE-Core rev:
a7d519f742aadc9110c2401f359254210a784f6b)

Quoted 7 lines
Quoted 6 lines
Quoted 4 lines
Quoted 1 line

Does warrior otherwise work on gcc10 hosts?

Not yet but I am working on it.

I think to make a decision on whether we plan to support gcc10 or not we'll need to know the scope of the changes. I'm starting to worry this would be a bit too invasive, particularly as warrior is about to enter community support or EOL depending on whether there is a maintainer.

Cheers,

Richard

#4 lore raw
Quoted 20 lines · Richard Purdie
Quoted 18 lines · Andreas Müller
Quoted 16 lines · Richard Purdie
Quoted 14 lines · Andreas Müller

From: Jacob Kroon <[email protected]>

'pseudo_access_t' is a type, so use typedef.

Fixes building pseudo with gcc 10 where -fno-common is the default.

(Backport of OE-Core rev:
a7d519f742aadc9110c2401f359254210a784f6b)

Quoted 1 line
Quoted 8 lines
Quoted 2 lines
Quoted 1 line

Does warrior otherwise work on gcc10 hosts?

Not yet but I am working on it.

I think to make a decision on whether we plan to support gcc10 or not we'll need to know the scope of the changes. I'm starting to worry this would be a bit too invasive, particularly as warrior is about to enter community support or EOL depending on whether there is a maintainer.

I see. Problems for me - job's hat on - are:

1. we're close to release a product based on warrior (and a new colleague of mine has just installed Fedora 32 with gcc 10 - that's why I ask for it)
2. my tests with dunfell were - hmm - not running out of the box exactly.

So we should decide to upgrade to dunfell. That is what I use for private raspi-images an am very close to happy.

Thanks for 'warning' me

Andreas

#5 lore raw
Quoted 32 lines · Andreas Müller
Quoted 20 lines · Richard Purdie
Quoted 18 lines · Andreas Müller
Quoted 16 lines · Richard Purdie
Quoted 14 lines · Andreas Müller

From: Jacob Kroon <[email protected]>

'pseudo_access_t' is a type, so use typedef.

Fixes building pseudo with gcc 10 where -fno-common is the default.

(Backport of OE-Core rev:
a7d519f742aadc9110c2401f359254210a784f6b)

Quoted 1 line
Quoted 8 lines
Quoted 2 lines
Quoted 1 line

Does warrior otherwise work on gcc10 hosts?

Not yet but I am working on it.

I think to make a decision on whether we plan to support gcc10 or not we'll need to know the scope of the changes. I'm starting to worry this would be a bit too invasive, particularly as warrior is about to enter community support or EOL depending on whether there is a maintainer.

I see. Problems for me - job's hat on - are:

Quoted 8 lines

1. we're close to release a product based on warrior (and a new colleague of mine has just installed Fedora 32 with gcc 10 - that's why I ask for it)
2. my tests with dunfell were - hmm - not running out of the box exactly.

So we should decide to upgrade to dunfell. That is what I use for private raspi-images an am very close to happy.

Dang, I was silently but closely following the discussion hoping to be able to apply more or less the same patches on top of thud. Same issue as yours, colleagues moving to Ubuntu 20.04 with gcc 10+. I guess it's docker time :/ (there'll always be that one client that will not want to upgrade to newer releases).

Or maybe we can support gcc10 on the community branch once it's EOL (or doing the job ourselves..). What seems to be the big worry about supporting gcc10 on warrior/thud? Just so I can explain why we should go the docker route instead of spending time on making thud (and we still have krogoth for some...) build with gcc10 and maybe start working on upgrading those we can to newer releases. Basically what are the expected big hurdles on the way?

Cheers,
Quentin

#6 lore raw
Quoted 46 lines · Quentin Schulz
Quoted 30 lines · Andreas Müller
Quoted 18 lines · Richard Purdie
Quoted 17 lines · Andreas Müller
Quoted 15 lines · Richard Purdie
Quoted 13 lines · Andreas Müller

From: Jacob Kroon <[email protected]>

'pseudo_access_t' is a type, so use typedef.

Fixes building pseudo with gcc 10 where -fno-common is the default.

(Backport of OE-Core rev:
a7d519f742aadc9110c2401f359254210a784f6b)

Quoted 8 lines
Quoted 2 lines
Quoted 1 line

Does warrior otherwise work on gcc10 hosts?

Not yet but I am working on it.

I think to make a decision on whether we plan to support gcc10 or not we'll need to know the scope of the changes. I'm starting to worry this would be a bit too invasive, particularly as warrior is about to enter community support or EOL depending on whether there is a maintainer.

I see. Problems for me - job's hat on - are:

Quoted 8 lines

1. we're close to release a product based on warrior (and a new colleague of mine has just installed Fedora 32 with gcc 10 - that's why I ask for it)
2. my tests with dunfell were - hmm - not running out of the box exactly.

So we should decide to upgrade to dunfell. That is what I use for private raspi-images an am very close to happy.

Dang, I was silently but closely following the discussion hoping to be able to apply more or less the same patches on top of thud. Same issue as yours, colleagues moving to Ubuntu 20.04 with gcc 10+. I guess it's docker time :/ (there'll always be that one client that will not want to upgrade to newer releases).

Quoted 7 lines

Or maybe we can support gcc10 on the community branch once it's EOL (or doing the job ourselves..). What seems to be the big worry about supporting gcc10 on warrior/thud? Just so I can explain why we should go the docker route instead of spending time on making thud (and we still have krogoth for some...) build with gcc10 and maybe start working on upgrading those we can to newer releases. Basically what are the expected big hurdles on the way?

Your other option would be buildtools tarball for the f32/ubuntu2004 systems.

I could see a case for backporting the buildtools tarball installer/wrapper for those older releases as a way of fixing these issues.

Cheers,

Richard

#7 lore raw

Ubuntu 20.04 doesn't have gcc 10 as default, it's still 9. So I guess that's different from Fedora 32 that someone mentioned.

Regards
//Ernst

Thu 2020-06-04 klockan 15:06 +0100 skrev Richard Purdie:

On Thu, 2020-06-04 at 15:57 +0200, Quentin Schulz wrote:

On Thu, Jun 04, 2020 at 03:49:28PM +0200, Andreas Müller wrote:

On Thu, Jun 4, 2020 at 2:47 PM Richard Purdie

<

<mailto:[email protected]>

[email protected]

Quoted 1 line

wrote:

On Thu, 2020-06-04 at 14:18 +0200, Andreas Müller wrote:

On Thu, Jun 4, 2020 at 2:11 PM Richard Purdie

<

<mailto:[email protected]>

[email protected]

Quoted 1 line

wrote:

On Thu, 2020-06-04 at 14:03 +0200, Andreas Müller wrote:

From: Jacob Kroon <

<mailto:[email protected]>

[email protected]

Quoted 1 line

'pseudo_access_t' is a type, so use typedef.

Fixes building pseudo with gcc 10 where -fno-common is the

default.

(Backport of OE-Core rev:

a7d519f742aadc9110c2401f359254210a784f6b)

Signed-off-by Jacob Kroon <

<mailto:[email protected]>

[email protected]

Quoted 1 line
Signed-off-by Richard Purdie <

<mailto:[email protected]>

[email protected]

Signed-off-by Andreas Müller <

<mailto:[email protected]>

[email protected]

Quoted 1 line

Does warrior otherwise work on gcc10 hosts?

Not yet but I am working on it.

I think to make a decision on whether we plan to support gcc10 or not

we'll need to know the scope of the changes. I'm starting to worry this

would be a bit too invasive, particularly as warrior is about to enter

community support or EOL depending on whether there is a maintainer.

I see. Problems for me - job's hat on - are:

1. we're close to release a product based on warrior (and a new

colleague of mine has just installed Fedora 32 with gcc 10 - that's

why I ask for it)

2. my tests with dunfell were - hmm - not running out of the box exactly.

So we should decide to upgrade to dunfell. That is what I use for

private raspi-images an am very close to happy.

Dang, I was silently but closely following the discussion hoping to be

able to apply more or less the same patches on top of thud. Same issue

as yours, colleagues moving to Ubuntu 20.04 with gcc 10+. I guess it's

docker time :/ (there'll always be that one client that will not want to

upgrade to newer releases).

Or maybe we can support gcc10 on the community branch once it's EOL (or

doing the job ourselves..). What seems to be the big worry about

supporting gcc10 on warrior/thud? Just so I can explain why we should go

the docker route instead of spending time on making thud (and we still

have krogoth for some...) build with gcc10 and maybe start working on

upgrading those we can to newer releases. Basically what are the

expected big hurdles on the way?

Your other option would be buildtools tarball for the f32/ubuntu2004

systems.

I could see a case for backporting the buildtools tarball

installer/wrapper for those older releases as a way of fixing these

issues.

Cheers,

Richard

#8 lore raw
In reply to Quentin Schulz

Maybe I'm reading RP's reply differently, but I think he just asked to collect all necessary changes for warrior to support newer host, before they are accepted.

It doesn't help much when only one smaller issue is fixed and the rest still needs to be figured out locally and the commit message might even mislead people in assuming that gcc-10 on host is now supported in warrior/thud.

FWIW: I've built some smaller thud based images on 20.04 with only small modifications, e.g. fix for qemu:
https://git.openembedded.org/openembedded-core-contrib/commit/?h=jansa/thud&id=68475d0255c4089538886d1ee2cab6b6efd62ce1
there aren't probably many issues left, so it just needs someone to test it more properly and collect all needed fixes to apply them at the same time or give up if it shows too many difficult to solve issues (I doubt that).

In worst case they might be rejected, but then it would be still useful to have them in one pull request and share them with other interested parties.

Cheers,

Quoted 61 lines · Quentin Schulz
Quoted 11 lines · Andreas Müller
Quoted 9 lines · Richard Purdie
Quoted 7 lines · Andreas Müller
Quoted 5 lines · Richard Purdie
Quoted 4 lines · Andreas Müller

From: Jacob Kroon <[email protected]>

'pseudo_access_t' is a type, so use typedef.

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines

Fixes building pseudo with gcc 10 where -fno-common is the default.

(Backport of OE-Core rev:
a7d519f742aadc9110c2401f359254210a784f6b)

Signed-off-by Richard Purdie <

[email protected]

Quoted 13 lines
Quoted 11 lines
Quoted 5 lines
Quoted 4 lines
Quoted 2 lines
Quoted 1 line

Does warrior otherwise work on gcc10 hosts?

Not yet but I am working on it.

I think to make a decision on whether we plan to support gcc10 or not we'll need to know the scope of the changes. I'm starting to worry this would be a bit too invasive, particularly as warrior is about to enter community support or EOL depending on whether there is a maintainer.

I see. Problems for me - job's hat on - are:

Quoted 8 lines

1. we're close to release a product based on warrior (and a new colleague of mine has just installed Fedora 32 with gcc 10 - that's why I ask for it)
2. my tests with dunfell were - hmm - not running out of the box exactly.

So we should decide to upgrade to dunfell. That is what I use for private raspi-images an am very close to happy.

Dang, I was silently but closely following the discussion hoping to be able to apply more or less the same patches on top of thud. Same issue as yours, colleagues moving to Ubuntu 20.04 with gcc 10+. I guess it's docker time :/ (there'll always be that one client that will not want to upgrade to newer releases).

Or maybe we can support gcc10 on the community branch once it's EOL (or doing the job ourselves..). What seems to be the big worry about supporting gcc10 on warrior/thud? Just so I can explain why we should go the docker route instead of spending time on making thud (and we still have krogoth for some...) build with gcc10 and maybe start working on upgrading those we can to newer releases. Basically what are the expected big hurdles on the way?

Cheers,
Quentin