Re: [oe] [meta-networking][PATCH][V3] netmap: fix multiple build issues.

[Joe MacDonald]

[Re: [meta-networking][PATCH][V3] netmap: fix multiple build issues.] On 
15.09.15 (Tue 07:15) akuster wrote:

> 
> 
> On 09/14/2015 06:49 PM, Joe MacDonald wrote:
> > Hi Armin,
> > 
> > [Re: [meta-networking][PATCH][V3] netmap: fix multiple build issues.] On 
> > 15.09.14 (Mon 14:23) akuster wrote:
> > 
> >> Joe,
> >>
> >> On 09/14/2015 12:51 PM, Joe MacDonald wrote:
> >>> Hey Armin,
> >>>
> >>> I'm not quite sure what happened here, but when I went to cherry-pick
> >>> this from master-next it appears to be patching the recipe introduced in
> >>> v2 of the original netmap submission, but I don't see that merged to
> >>> master yet and I didn't see a follow-up to the email I sent about the v2
> >>> patch here:
> >>>
> >>>http://patchwork.openembedded.org/patch/99753/
> >>>
> >>> Curiously, I also don't see my comments in the mailing list archive, so
> >>> it might be that somehow my mail got discarded from the list even though
> >>> patchwork saw it (and I know you saw it because you sent me a follow-up
> >>> to my follow-up, even).
> >>>
> >>> I can bring the both of them in, but frankly if the two commits together
> >>> (ae7fed356adf3bab37822230290f3aff7390d75c and
> >>> 0a86d95779b64b4b1a0f91898616217006ace16c in master-next) introduce a new
> >>> recipe that builds successfully, I'd rather take it as a single commit
> >>> in master (which would also give you another shot at removing the 'V2: '
> >>> text from the commit log that Martin mentioned ...)
> >>
> >> So would a clean  patch adding the package make sense at this point? My
> >> last V3 was on top of what is in master-next. ( guesses that is not
> >> appropriate process). ie one against master not master-next?
> >>
> >>
> >>>
> >>> Thoughts?  I've not yet changed the status of this patch in patchwork
> >>> and probably won't until I hear back from you.  Let's call it 'pending'.
> >>> :-)
> >>
> >> I will do what ever is easiest for you.
> > 
> > Since nothing's been merged from master-next into master yet, I'd rather
> > just have a single patch that combines the two you have out right now.
> > I was going to squash them together myself, but then thought I should
> > check with you first just in case there was something else I was
> > missing.
> 
> Nope those are it. The one in Master-next and the [V3] one.

Okay, I'll take care of that, then, since it seems to have sorted things
out.  Thanks Armin.

-J.

> 
> regards,
> Armin
> > 
> > -J.
> > 
> >>
> >> - armin
> >>>
> >>> -J.
> >>>
> >>> [[meta-networking][PATCH][V3] netmap: fix multiple build issues.] On 
> >>> 15.09.08 (Tue 14:32) Armin Kuster wrote:
> >>>
>  From: Armin Kuster 
> 
>  updated to latest version to get kernel 4.1 support
> 
>  update makefile patch
>  fixed printf type issue
>  Fixed manual config options
>  Readme checksum changed, don't know why. must got it wrong to begin with.
> 
>  V2: Fix "Upstream-Status" typos
> 
>  Signed-off-by: Armin Kuster 
>  ---
>   ...1-testmmap-fix-compile-issue-with-gcc-5.x.patch | 33 
>  +
>   .../netmap/files/makefile_fixup.patch  | 34 
>  +-
>   .../recipes-kernel/netmap/netmap-modules_git.bb|  8 -
>   meta-networking/recipes-kernel/netmap/netmap.inc   |  4 +--
>   .../recipes-kernel/netmap/netmap_git.bb|  1 +
>   5 files changed, 69 insertions(+), 11 deletions(-)
>   create mode 100644 
>  meta-networking/recipes-kernel/netmap/files/0001-testmmap-fix-compile-issue-with-gcc-5.x.patch
> 
>  diff --git 
>  a/meta-networking/recipes-kernel/netmap/files/0001-testmmap-fix-compile-issue-with-gcc-5.x.patch
>   
>  b/meta-networking/recipes-kernel/netmap/files/0001-testmmap-fix-compile-issue-with-gcc-5.x.patch
>  new file mode 100644
>  index 000..ea36b1c
>  --- /dev/null
>  +++ 
>  b/meta-networking/recipes-kernel/netmap/files/0001-testmmap-fix-compile-issue-with-gcc-5.x.patch
>  @@ -0,0 +1,33 @@
>  +Upstream-Status: Pending
>  +
>  +From c81bf54d6eb870286662a11d3b4a994717c47696 Mon Sep 17 00:00:00 2001
>  +From: Armin Kuster 
>  +Date: Tue, 8 Sep 2015 05:36:27 -0700
>  +Subject: [PATCH] testmmap: fix compile issue with gcc 5.x
>  +
>  +this fixes:
>  +examples/testmmap.c:540:10: error: format '%ld' expects argument of 
>  type 'long int', but argument 3 has type 'ssize_t {aka const int}' 
>  [-Werror=format=]
>  +|printf("ring_ofs[%d] %ld\n", i, nifp->ring_ofs[i]);
>  +|   ^
>  +
>  +Signed-off-by: Armin Kuster 
>  +---
>  + examples/testmmap.c | 2 +-
>  + 1 file changed, 1 insertion(+), 1 deletion(-)
>  +
>  +diff --git a/examples/testmmap.c b/examples/testmmap.c
>  +index d7f6acc..934489a 100644
>  +--- a/examples/testmmap.c
>   

Re: [oe] [meta-networking][PATCH][V3] netmap: fix multiple build issues.

[Joe MacDonald]

Hi Armin,

[Re: [meta-networking][PATCH][V3] netmap: fix multiple build issues.] On 
15.09.14 (Mon 14:23) akuster wrote:

> Joe,
> 
> On 09/14/2015 12:51 PM, Joe MacDonald wrote:
> > Hey Armin,
> > 
> > I'm not quite sure what happened here, but when I went to cherry-pick
> > this from master-next it appears to be patching the recipe introduced in
> > v2 of the original netmap submission, but I don't see that merged to
> > master yet and I didn't see a follow-up to the email I sent about the v2
> > patch here:
> > 
> >http://patchwork.openembedded.org/patch/99753/
> > 
> > Curiously, I also don't see my comments in the mailing list archive, so
> > it might be that somehow my mail got discarded from the list even though
> > patchwork saw it (and I know you saw it because you sent me a follow-up
> > to my follow-up, even).
> > 
> > I can bring the both of them in, but frankly if the two commits together
> > (ae7fed356adf3bab37822230290f3aff7390d75c and
> > 0a86d95779b64b4b1a0f91898616217006ace16c in master-next) introduce a new
> > recipe that builds successfully, I'd rather take it as a single commit
> > in master (which would also give you another shot at removing the 'V2: '
> > text from the commit log that Martin mentioned ...)
> 
> So would a clean  patch adding the package make sense at this point? My
> last V3 was on top of what is in master-next. ( guesses that is not
> appropriate process). ie one against master not master-next?
> 
> 
> > 
> > Thoughts?  I've not yet changed the status of this patch in patchwork
> > and probably won't until I hear back from you.  Let's call it 'pending'.
> > :-)
> 
> I will do what ever is easiest for you.

Since nothing's been merged from master-next into master yet, I'd rather
just have a single patch that combines the two you have out right now.
I was going to squash them together myself, but then thought I should
check with you first just in case there was something else I was
missing.

-J.

> 
> - armin
> > 
> > -J.
> > 
> > [[meta-networking][PATCH][V3] netmap: fix multiple build issues.] On 
> > 15.09.08 (Tue 14:32) Armin Kuster wrote:
> > 
> >> From: Armin Kuster 
> >>
> >> updated to latest version to get kernel 4.1 support
> >>
> >> update makefile patch
> >> fixed printf type issue
> >> Fixed manual config options
> >> Readme checksum changed, don't know why. must got it wrong to begin with.
> >>
> >> V2: Fix "Upstream-Status" typos
> >>
> >> Signed-off-by: Armin Kuster 
> >> ---
> >>  ...1-testmmap-fix-compile-issue-with-gcc-5.x.patch | 33 
> >> +
> >>  .../netmap/files/makefile_fixup.patch  | 34 
> >> +-
> >>  .../recipes-kernel/netmap/netmap-modules_git.bb|  8 -
> >>  meta-networking/recipes-kernel/netmap/netmap.inc   |  4 +--
> >>  .../recipes-kernel/netmap/netmap_git.bb|  1 +
> >>  5 files changed, 69 insertions(+), 11 deletions(-)
> >>  create mode 100644 
> >> meta-networking/recipes-kernel/netmap/files/0001-testmmap-fix-compile-issue-with-gcc-5.x.patch
> >>
> >> diff --git 
> >> a/meta-networking/recipes-kernel/netmap/files/0001-testmmap-fix-compile-issue-with-gcc-5.x.patch
> >>  
> >> b/meta-networking/recipes-kernel/netmap/files/0001-testmmap-fix-compile-issue-with-gcc-5.x.patch
> >> new file mode 100644
> >> index 000..ea36b1c
> >> --- /dev/null
> >> +++ 
> >> b/meta-networking/recipes-kernel/netmap/files/0001-testmmap-fix-compile-issue-with-gcc-5.x.patch
> >> @@ -0,0 +1,33 @@
> >> +Upstream-Status: Pending
> >> +
> >> +From c81bf54d6eb870286662a11d3b4a994717c47696 Mon Sep 17 00:00:00 2001
> >> +From: Armin Kuster 
> >> +Date: Tue, 8 Sep 2015 05:36:27 -0700
> >> +Subject: [PATCH] testmmap: fix compile issue with gcc 5.x
> >> +
> >> +this fixes:
> >> +examples/testmmap.c:540:10: error: format '%ld' expects argument of type 
> >> 'long int', but argument 3 has type 'ssize_t {aka const int}' 
> >> [-Werror=format=]
> >> +|printf("ring_ofs[%d] %ld\n", i, nifp->ring_ofs[i]);
> >> +|   ^
> >> +
> >> +Signed-off-by: Armin Kuster 
> >> +---
> >> + examples/testmmap.c | 2 +-
> >> + 1 file changed, 1 insertion(+), 1 deletion(-)
> >> +
> >> +diff --git a/examples/testmmap.c b/examples/testmmap.c
> >> +index d7f6acc..934489a 100644
> >> +--- a/examples/testmmap.c
> >>  b/examples/testmmap.c
> >> +@@ -537,7 +537,7 @@ do_if()
> >> +  for (i = 0; i < 5; i++)
> >> +  printf("spare1[%d]  %u\n", i, nifp->ni_spare1[i]);
> >> +  for (i = 0; i < (nifp->ni_tx_rings + nifp->ni_rx_rings + 2); i++)
> >> +- printf("ring_ofs[%d] %ld\n", i, nifp->ring_ofs[i]);
> >> ++ printf("ring_ofs[%d] %zd\n", i, nifp->ring_ofs[i]);
> >> + }
> >> + 
> >> + struct netmap_ring *
> >> +-- 
> >> +2.3.5
> >> +
> >> diff --git 
> >> a/meta-networking/recipes-kernel/netmap/files/makefile_fixup.patch 
> >> b/meta-networking/recipes-kernel/netmap/files/makefile_fixup.patch
> >> index 

Re: [oe] [meta-networking][PATCH][V3] netmap: fix multiple build issues.

[Joe MacDonald]

Hey Armin,

I'm not quite sure what happened here, but when I went to cherry-pick
this from master-next it appears to be patching the recipe introduced in
v2 of the original netmap submission, but I don't see that merged to
master yet and I didn't see a follow-up to the email I sent about the v2
patch here:

   http://patchwork.openembedded.org/patch/99753/

Curiously, I also don't see my comments in the mailing list archive, so
it might be that somehow my mail got discarded from the list even though
patchwork saw it (and I know you saw it because you sent me a follow-up
to my follow-up, even).

I can bring the both of them in, but frankly if the two commits together
(ae7fed356adf3bab37822230290f3aff7390d75c and
0a86d95779b64b4b1a0f91898616217006ace16c in master-next) introduce a new
recipe that builds successfully, I'd rather take it as a single commit
in master (which would also give you another shot at removing the 'V2: '
text from the commit log that Martin mentioned ...)

Thoughts?  I've not yet changed the status of this patch in patchwork
and probably won't until I hear back from you.  Let's call it 'pending'.
:-)

-J.

[[meta-networking][PATCH][V3] netmap: fix multiple build issues.] On 15.09.08 
(Tue 14:32) Armin Kuster wrote:

> From: Armin Kuster 
> 
> updated to latest version to get kernel 4.1 support
> 
> update makefile patch
> fixed printf type issue
> Fixed manual config options
> Readme checksum changed, don't know why. must got it wrong to begin with.
> 
> V2: Fix "Upstream-Status" typos
> 
> Signed-off-by: Armin Kuster 
> ---
>  ...1-testmmap-fix-compile-issue-with-gcc-5.x.patch | 33 +
>  .../netmap/files/makefile_fixup.patch  | 34 
> +-
>  .../recipes-kernel/netmap/netmap-modules_git.bb|  8 -
>  meta-networking/recipes-kernel/netmap/netmap.inc   |  4 +--
>  .../recipes-kernel/netmap/netmap_git.bb|  1 +
>  5 files changed, 69 insertions(+), 11 deletions(-)
>  create mode 100644 
> meta-networking/recipes-kernel/netmap/files/0001-testmmap-fix-compile-issue-with-gcc-5.x.patch
> 
> diff --git 
> a/meta-networking/recipes-kernel/netmap/files/0001-testmmap-fix-compile-issue-with-gcc-5.x.patch
>  
> b/meta-networking/recipes-kernel/netmap/files/0001-testmmap-fix-compile-issue-with-gcc-5.x.patch
> new file mode 100644
> index 000..ea36b1c
> --- /dev/null
> +++ 
> b/meta-networking/recipes-kernel/netmap/files/0001-testmmap-fix-compile-issue-with-gcc-5.x.patch
> @@ -0,0 +1,33 @@
> +Upstream-Status: Pending
> +
> +From c81bf54d6eb870286662a11d3b4a994717c47696 Mon Sep 17 00:00:00 2001
> +From: Armin Kuster 
> +Date: Tue, 8 Sep 2015 05:36:27 -0700
> +Subject: [PATCH] testmmap: fix compile issue with gcc 5.x
> +
> +this fixes:
> +examples/testmmap.c:540:10: error: format '%ld' expects argument of type 
> 'long int', but argument 3 has type 'ssize_t {aka const int}' 
> [-Werror=format=]
> +|printf("ring_ofs[%d] %ld\n", i, nifp->ring_ofs[i]);
> +|   ^
> +
> +Signed-off-by: Armin Kuster 
> +---
> + examples/testmmap.c | 2 +-
> + 1 file changed, 1 insertion(+), 1 deletion(-)
> +
> +diff --git a/examples/testmmap.c b/examples/testmmap.c
> +index d7f6acc..934489a 100644
> +--- a/examples/testmmap.c
>  b/examples/testmmap.c
> +@@ -537,7 +537,7 @@ do_if()
> + for (i = 0; i < 5; i++)
> + printf("spare1[%d]  %u\n", i, nifp->ni_spare1[i]);
> + for (i = 0; i < (nifp->ni_tx_rings + nifp->ni_rx_rings + 2); i++)
> +-printf("ring_ofs[%d] %ld\n", i, nifp->ring_ofs[i]);
> ++printf("ring_ofs[%d] %zd\n", i, nifp->ring_ofs[i]);
> + }
> + 
> + struct netmap_ring *
> +-- 
> +2.3.5
> +
> diff --git a/meta-networking/recipes-kernel/netmap/files/makefile_fixup.patch 
> b/meta-networking/recipes-kernel/netmap/files/makefile_fixup.patch
> index e838653..b050bf0 100644
> --- a/meta-networking/recipes-kernel/netmap/files/makefile_fixup.patch
> +++ b/meta-networking/recipes-kernel/netmap/files/makefile_fixup.patch
> @@ -10,24 +10,24 @@ Index: LINUX/netmap.mak.in
>  ===
>  --- a/LINUX/netmap.mak.in
>  +++ b/LINUX/netmap.mak.in
> -@@ -12,10 +12,8 @@ SRCDIR:=@SRCDIR@
> +@@ -12,10 +12,9 @@ SRCDIR:=@SRCDIR@
>   # The following commands are needed to build the modules as out-of-tree,
>   # in fact the kernel sources path must be specified.
>   
>  -PWD ?= $(CURDIR)
> --
> + 
>   # Additional compile flags (e.g. header location)
>  -EXTRA_CFLAGS := -I$(PWD) -I$(SRCDIR) -I$(SRCDIR)/../sys 
> -I$(SRCDIR)/../sys/dev -DCONFIG_NETMAP
>  +EXTRA_CFLAGS := -I$(SRCDIR) -I$(SRCDIR)/../sys -I$(SRCDIR)/../sys/dev 
> -DCONFIG_NETMAP
>   EXTRA_CFLAGS += -Wno-unused-but-set-variable
> - EXTRA_CFLAGS += $(foreach s,$(SUBSYS),-DCONFIG_NETMAP_$(shell echo $s|tr 
> a-z A-Z))
> + EXTRA_CFLAGS += $(foreach s,$(SUBSYS),-DCONFIG_NETMAP_$(shell echo $s|tr 
> a-z- A-Z_))
>   
>  

[oe] [meta-networking][PATCH][V3] netmap: fix multiple build issues.

[Armin Kuster]

From: Armin Kuster 

updated to latest version to get kernel 4.1 support

update makefile patch
fixed printf type issue
Fixed manual config options
Readme checksum changed, don't know why. must got it wrong to begin with.

V2: Fix "Upstream-Status" typos

Signed-off-by: Armin Kuster 
---
 ...1-testmmap-fix-compile-issue-with-gcc-5.x.patch | 33 +
 .../netmap/files/makefile_fixup.patch  | 34 +-
 .../recipes-kernel/netmap/netmap-modules_git.bb|  8 -
 meta-networking/recipes-kernel/netmap/netmap.inc   |  4 +--
 .../recipes-kernel/netmap/netmap_git.bb|  1 +
 5 files changed, 69 insertions(+), 11 deletions(-)
 create mode 100644 
meta-networking/recipes-kernel/netmap/files/0001-testmmap-fix-compile-issue-with-gcc-5.x.patch

diff --git 
a/meta-networking/recipes-kernel/netmap/files/0001-testmmap-fix-compile-issue-with-gcc-5.x.patch
 
b/meta-networking/recipes-kernel/netmap/files/0001-testmmap-fix-compile-issue-with-gcc-5.x.patch
new file mode 100644
index 000..ea36b1c
--- /dev/null
+++ 
b/meta-networking/recipes-kernel/netmap/files/0001-testmmap-fix-compile-issue-with-gcc-5.x.patch
@@ -0,0 +1,33 @@
+Upstream-Status: Pending
+
+From c81bf54d6eb870286662a11d3b4a994717c47696 Mon Sep 17 00:00:00 2001
+From: Armin Kuster 
+Date: Tue, 8 Sep 2015 05:36:27 -0700
+Subject: [PATCH] testmmap: fix compile issue with gcc 5.x
+
+this fixes:
+examples/testmmap.c:540:10: error: format '%ld' expects argument of type 'long 
int', but argument 3 has type 'ssize_t {aka const int}' [-Werror=format=]
+|printf("ring_ofs[%d] %ld\n", i, nifp->ring_ofs[i]);
+|   ^
+
+Signed-off-by: Armin Kuster 
+---
+ examples/testmmap.c | 2 +-
+ 1 file changed, 1 insertion(+), 1 deletion(-)
+
+diff --git a/examples/testmmap.c b/examples/testmmap.c
+index d7f6acc..934489a 100644
+--- a/examples/testmmap.c
 b/examples/testmmap.c
+@@ -537,7 +537,7 @@ do_if()
+   for (i = 0; i < 5; i++)
+   printf("spare1[%d]  %u\n", i, nifp->ni_spare1[i]);
+   for (i = 0; i < (nifp->ni_tx_rings + nifp->ni_rx_rings + 2); i++)
+-  printf("ring_ofs[%d] %ld\n", i, nifp->ring_ofs[i]);
++  printf("ring_ofs[%d] %zd\n", i, nifp->ring_ofs[i]);
+ }
+ 
+ struct netmap_ring *
+-- 
+2.3.5
+
diff --git a/meta-networking/recipes-kernel/netmap/files/makefile_fixup.patch 
b/meta-networking/recipes-kernel/netmap/files/makefile_fixup.patch
index e838653..b050bf0 100644
--- a/meta-networking/recipes-kernel/netmap/files/makefile_fixup.patch
+++ b/meta-networking/recipes-kernel/netmap/files/makefile_fixup.patch
@@ -10,24 +10,24 @@ Index: LINUX/netmap.mak.in
 ===
 --- a/LINUX/netmap.mak.in
 +++ b/LINUX/netmap.mak.in
-@@ -12,10 +12,8 @@ SRCDIR:=@SRCDIR@
+@@ -12,10 +12,9 @@ SRCDIR:=@SRCDIR@
  # The following commands are needed to build the modules as out-of-tree,
  # in fact the kernel sources path must be specified.
  
 -PWD ?= $(CURDIR)
--
+ 
  # Additional compile flags (e.g. header location)
 -EXTRA_CFLAGS := -I$(PWD) -I$(SRCDIR) -I$(SRCDIR)/../sys 
-I$(SRCDIR)/../sys/dev -DCONFIG_NETMAP
 +EXTRA_CFLAGS := -I$(SRCDIR) -I$(SRCDIR)/../sys -I$(SRCDIR)/../sys/dev 
-DCONFIG_NETMAP
  EXTRA_CFLAGS += -Wno-unused-but-set-variable
- EXTRA_CFLAGS += $(foreach s,$(SUBSYS),-DCONFIG_NETMAP_$(shell echo $s|tr a-z 
A-Z))
+ EXTRA_CFLAGS += $(foreach s,$(SUBSYS),-DCONFIG_NETMAP_$(shell echo $s|tr a-z- 
A-Z_))
  
 Index: LINUX/configure
 ===
 --- a/LINUX/configure
 +++ b/LINUX/configure
-@@ -311,34 +311,6 @@ reset_tests() {
-   NEXTTEST=1
+@@ -349,52 +349,6 @@ reset_tests() {
+   EOF
  }
  
 -# run_tests: run all accumulated tests and exec the pertinent
@@ -48,28 +48,46 @@ Index: LINUX/configure
 -  cat >> $TMPDIR/Makefile <<-EOF
 -  get-$d:
 -  $t  [ -z "\$($d-src)" ] || cp -Rp \$($d-src) \$(if 
\$($d-dst),\$($d-dst),.)
+-  $t  touch get-$d
 -  EOF
 -  done
 -  echo endif >> $TMPDIR/Makefile
+-  {
+-  cat <<-EOF
+-##
+-## BEGIN RUNNING TESTS: $(date)
+-##
+-## Makefile:
+-  EOF
+-  cat $TMPDIR/Makefile
+-  cat <<-EOF
+-##
+-  EOF
+-  } >> config.log
 -  (
 -  cd $TMPDIR
 -  make -k -j $(grep -c processor /proc/cpuinfo)
 -  ) >> config.log
 -  eval "$TESTPOSTPROC"
+-  cat >> config.log <<-EOF
+-##
+-## END RUNNING TESTS: $(date)