On Wed, Dec 2, 2015 at 6:14 AM, Bastian Bittorf wrote:
> * Petr Štetiar [30.11.2015 13:47]:
>> That sounds really great, can you please share more details? It's just your
>> guess or you've achieved such build times already?
>
> today tested PowerPC/mpc85xx and this needs 160 seconds:
> make clea
On Thu, Aug 20, 2015 at 6:57 AM, Jonas Gorski wrote:
> Hi,
>
> On Thu, Aug 20, 2015 at 11:40 AM, Dirk Neukirchen
> wrote:
>> GCC 4.9 was marked BROKEN with r46089. Recently GCC 5.2 was added (marked as
>> broken too)
>>
>> commit msg: " toolchain: mark gcc 4.9 as broken, it miscompiles some cod
On Mon, Mar 30, 2015 at 3:21 PM, John Crispin wrote:
[snip]
> fix in the making, will be pushed tomorrow ...
Thanks John!
-John
___
openwrt-devel mailing list
openwrt-devel@lists.openwrt.org
https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-de
Commit 988e3615e52b2cd9b9e07f4bbbd76494f7e0cb24 broke building
e2fsprogs on an ARM platform. It looks like that in version 1.42.10
the decision was made to drop locally defined versions of both
sync_file_range() and fallocate() because they were broken on 32-bit
platforms, and there was some issue
On Mon, Mar 23, 2015 at 10:50 AM, Jo-Philipp Wich wrote:
> Hi.
>
>> Thanks, I got it now. So I should run make oldconfig when compiling
>> automatically everyday?
>
> The best approach is the following:
>
> - Initially use menuconfig and select stuff as you like
> - Extract the delta compared to t
There are some places where there is a redundant declaration of
strlcpy() that prevents building perf otherwise.
Signed-off-by: John Szakmeister
---
package/devel/perf/Makefile | 1 +
1 file changed, 1 insertion(+)
diff --git a/package/devel/perf/Makefile b/package/devel/perf/Makefile
index
Signed-off-by: John Szakmeister
---
target/linux/generic/config-3.14 | 1 +
target/linux/generic/config-3.18 | 1 +
target/linux/generic/config-3.19 | 1 +
target/linux/generic/config-4.0 | 1 +
4 files changed, 4 insertions(+)
diff --git a/target/linux/generic/config-3.14 b/target/linux
On Mon, Feb 2, 2015 at 9:07 AM, Sławomir Demeszko
wrote:
[snip]
> I just tested and found that also a dot in filenames can be a problem. For
> example if we would have "some.file.so" then it matches also "some2file.so"
> and it could not detect missing dependency. So -qxF is a more than bit
> bett
On Mon, Feb 2, 2015 at 6:24 AM, Sławomir Demeszko
wrote:
> Compilation of packages written in C++, like smartmontools, unrar, etc.
> fail with message:
>
> Package smartmontools is missing dependencies for the following libraries:
> libuClibc++.so.0
>
> This is due unescaped "++" in argument for g
On Fri, Jan 16, 2015 at 4:42 AM, Rafał Miłecki wrote:
[snip]
> - if (uh->state == HTTP_STATE_RECV_DATA && uc->cb->data_read)
> - uc->cb->data_read(uc);
> + if (uh->state == HTTP_STATE_RECV_DATA)
> + /* Now it's uclient user turn to read some data */
> +
On Sat, Dec 20, 2014 at 9:32 AM, Felix Fietkau wrote:
> On 2014-12-20 14:49, John Szakmeister wrote:
>> Without this, modprobe doesn't work with uncompressed kernel modules.
>>
>> Signed-off-by: John Szakmeister
> Why do we need this patch in OpenWrt? We're not
Without this, modprobe doesn't work with uncompressed kernel modules.
Signed-off-by: John Szakmeister
---
.../008-upstream_no_fail_uncompressed.patch| 31 ++
1 file changed, 31 insertions(+)
create mode 100644
package/utils/busybox/patche
On Sat, Dec 13, 2014 at 6:08 AM, Felix Fietkau wrote:
[snip]
> How about this patch?
>
> diff --git a/toolchain/Makefile b/toolchain/Makefile
> index 36c6ed3..7f8d62f 100644
> --- a/toolchain/Makefile
> +++ b/toolchain/Makefile
> @@ -74,7 +74,7 @@ endif
>
> # prerequisites for the individual targ
On Sat, Dec 13, 2014 at 5:11 AM, Felix Fietkau wrote:
> On 2014-12-13 10:43, John Szakmeister wrote:
>> On Sat, Dec 13, 2014 at 4:28 AM, John Szakmeister
>> wrote:
>> [snip]
>>>> This doesn't look right to me, I don't think we should add the
>>&g
On Sat, Dec 13, 2014 at 4:28 AM, John Szakmeister wrote:
[snip]
>> This doesn't look right to me, I don't think we should add the
>> toolchain/prepare step as an intermediate target for the regular build
>> process.
>> How about just adding this line to
On Fri, Dec 12, 2014 at 12:03 PM, Felix Fietkau wrote:
> On 2014-12-09 11:47, John Szakmeister wrote:
>> This fixes an issue where the toolchain/prepare step could run, but some
>> of the necessary host tools might be missing.
>>
>> Signed-off-by: John Szakmeister
>
On Tue, Dec 9, 2014 at 5:46 AM, John Szakmeister wrote:
> Signed-off-by: John Szakmeister
> ---
My apologies, this should not be marked as a resend. It's new.
-John
___
openwrt-devel mailing list
openwrt-devel@lists.openwr
This fixes an issue where the toolchain/prepare step could run, but some
of the necessary host tools might be missing.
Signed-off-by: John Szakmeister
---
This is a resend of a patch I sent earlier
(https://lists.openwrt.org/pipermail/openwrt-devel/2014-October/028422.html).
I didn't receiv
Signed-off-by: John Szakmeister
---
This is a resend of a patch I sent earlier
(https://lists.openwrt.org/pipermail/openwrt-devel/2014-September/028069.html).
package/devel/gdb/Makefile | 1 +
1 file changed, 1 insertion(+)
diff --git a/package/devel/gdb/Makefile b/package/devel/gdb/Makefile
'll simply filter out stamp-bits from the symlinks
to be generated.
Note: gcc 4.4.7 is unaffected by this issue, so no fix is necessary
there.
Signed-off-by: John Szakmeister
---
This is a resend of a patch I sent earlier
(https://lists.openwrt.org/pipermail/openwrt-devel/2014-September/028
Signed-off-by: John Szakmeister
---
target/linux/at91/modules.mk | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/target/linux/at91/modules.mk b/target/linux/at91/modules.mk
index 4678b92..7f98233 100644
--- a/target/linux/at91/modules.mk
+++ b/target/linux/at91/modules.mk
On Mon, Nov 24, 2014 at 4:42 PM, Maxime Ripard
wrote:
> Rely on the Kconfig defconfig mechanism to fill all the missing options,
> instead of needing to set them all in the kernel configurations like what was
> previously done.
>
> This will allow to trim down a lot the configuration files, avoid
On Tue, Oct 14, 2014 at 6:59 AM, John Crispin wrote:
> these should really go into the at91 config file. the CONFIG_AT91_*
> ones for sure but most likely also the others.
So I tracked down the real problem. It's in configs/Config-kernel.in:
,[ Config-kernel.in ]-
| [...]
| 72 config KERNEL
---
Without these, my earlyprintk configuration was being lost.
-John
target/linux/generic/config-3.14 | 5 +
1 file changed, 5 insertions(+)
diff --git a/target/linux/generic/config-3.14 b/target/linux/generic/config-3.14
index 830efff..5f16281 100644
--- a/target/linux/generic/config-3.14
Should the following in a clean checkout (with a config) expect to
build correctly?
make package/kernel/mac80211/compile
I assumed it would, but was surprised when it did not. It didn't make
sure the tools and toolchain were built and ready, nor did it ensure
the kernel was in a good state f
This fixes an issue where the toolchain/prepare step could run, but some
of the necessary host tools might be missing.
Signed-off-by: John Szakmeister
---
This is an updated version of a patch I sent earlier. I believe
this fix to be more "proper", but I'd appreciate any feed
On Wed, Sep 17, 2014 at 3:00 PM, John Szakmeister wrote:
> This fixes an issue where the toolchain/prepare step could run, but some
> of the necessary host tools might be missing.
This is just a friendly follow up for this and the other patches. Is
there anything else I need to do? Is
This fixes an issue where the toolchain/prepare step could run, but some
of the necessary host tools might be missing.
Signed-off-by: John Szakmeister
---
I found that on two systems I could reliably make the build fail
with multiple jobs. It turns out that some of the host tools
being built
Signed-off-by: John Szakmeister
---
This doesn't necessarily fix anything that's "broken", but it's
nice to actually be able to read the lines instead of them
overwriting each other.
include/scan.mk | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/
Signed-off-by: John Szakmeister
---
package/devel/gdb/Makefile | 1 +
1 file changed, 1 insertion(+)
diff --git a/package/devel/gdb/Makefile b/package/devel/gdb/Makefile
index 7663188..9038115 100644
--- a/package/devel/gdb/Makefile
+++ b/package/devel/gdb/Makefile
@@ -42,6 +42,7 @@ endef
'll simply filter out stamp-bits from the symlinks
to be generated.
Signed-off-by: John Szakmeister
---
.../gcc/patches/4.6-linaro/930-no-clobber-stamp-bits.patch| 11 +++
.../gcc/patches/4.8-linaro/930-no-clobber-stamp-bits.patch| 11 +++
toolchain/gcc/patches/4.8.0
On Wed, Sep 17, 2014 at 1:40 AM, Jonathan Bennett wrote:
[snip]
> Now, what I think is happening here is that we have a race between a couple
> of compilation threads, one of which is the sed utility. What happens is
> something that uses sed is trying to compile before sed has finished
> compilin
On Wed, Sep 17, 2014 at 1:40 AM, Jonathan Bennett wrote:
> I'll throw a couple thoughts out, and then defer to someone that understands
> the openwrt build env better than I do.
> First, you do have sed installed on your host system, right? I don't think
> this is the issue, but best to check the
On Tue, Sep 16, 2014 at 9:32 PM, Jonathan Bennett wrote:
> In your tree, in the file tools/sed/Makefile, change HOST_BUILD_PARALLEL:=1
> to HOST_BUILD_PARALLEL:=0 and see if that fixes the error.
Unfortunately, it didn't help. I think the issue is that
linux-headers needs the SED utility, but it
On Tue, Sep 16, 2014 at 8:32 PM, Jonathan Bennett wrote:
> Are you building with multiple processes (-j 2 or greater)? This reminds me
> of an old bug that only popped up after a make distclean when building with
> multiple build threads.
Yes sir, I am. On the machine this is failing on, I'm att
I'm finding that I'm running into an issue occasionally when trying to
build. It appears that Make is trying to build gcc, but the host sed
hasn't been built yet so it ultimately fails:
checking for i686-linux-gnu-gcc option to accept ISO C89... bash:
/home/jszakmeister/projects/openwrt/staging_d
Signed-off-by: John Szakmeister
---
Using make with multiple jobs fails without this.
tools/Makefile | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/tools/Makefile b/tools/Makefile
index 09a0c8c..5629251 100644
--- a/tools/Makefile
+++ b/tools/Makefile
@@ -133,7 +133,7
On Tue, Mar 4, 2014 at 10:25 AM, Imre Kaloz wrote:
[snip]
> Generic (even arch specific) option go into
> target/linux/generic/config-$VERSION, and target specific options /
> overrides go into target/linux/$BOARD/config-{default,$VERSION}
>
> Subtarget config options override target config option
I'm looking to add support for a board into OpenWRT, and I'm curious
you guys so neatly split the configuration between the top-level, the
architecture, and the boards. It's an ARM-based board and we plan on
using device tree. With all the interaction between kernel
selections, I want to make sur
*bump*
-John
On Wed, Jul 3, 2013 at 3:29 PM, John Szakmeister wrote:
> Signed-off-by: John Szakmeister
> ---
> include/scan.mk | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> I noticed that the progress output was stepping all over itself, which made it
> har
On Mon, Sep 2, 2013 at 2:26 PM, Florian Fainelli wrote:
> Le mercredi 3 juillet 2013 15:32:21 John Szakmeister a écrit :
>> Users may have merge.ff set to false, which will force merge commits to
>> be generated. The intent here is to have a fast-forward merge when
>> po
lldpd 0.7.1 has several alignment issues that trip a system configured
to fault on misaligned accesses. Version 0.7.5 fixes that along with a
few other minor issues.
Signed-off-by: John Szakmeister
---
package/network/services/lldpd/Makefile | 4 ++--
1 file changed, 2 insertions(+), 2
Users may have merge.ff set to false, which will force merge commits to
be generated. The intent here is to have a fast-forward merge when
possible, so let's make sure fast-forwards are enabled.
Signed-off-by: John Szakmeister
---
scripts/feeds | 2 +-
1 file changed, 1 insertion(
Signed-off-by: John Szakmeister
---
include/scan.mk | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
I noticed that the progress output was stepping all over itself, which made it
hard to read. This uses a standard escape sequence to clear the line first.
-John
diff --git a/include
e the link without dereferencing any
symlinks by using the -n option.
Signed-off-by: John Szakmeister
---
Here's the patch with my Signed-off-by. Thank you for the feedback Florian!
-John
toolchain/Makefile | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/toolchain/
On Thu, Feb 28, 2013 at 5:03 AM, John Szakmeister wrote:
> It turns out that the symlink may exists already by the time the
> toolchain Makefile gets an opportunity to run. While we asked to
> replace the target, ln doesn't do so until after dereferencing the
> existing symlink.
It turns out that the symlink may exists already by the time the
toolchain Makefile gets an opportunity to run. While we asked to
replace the target, ln doesn't do so until after dereferencing the
existing symlink. This results in an unintended symlink that refers to
itself. Instead, create the
On Mon, Jan 7, 2013 at 10:51 AM, Oliver
wrote:
> On Monday 07 January 2013 13:17:24 Bastian Bittorf wrote:
>> * Imre Kaloz [07.01.2013 12:55]:
>> > >the problem is: most routers are "in production" and not "serial
>> > >attached" - so you have really a problem if a node ("rooftop" or "solar
>> >
48 matches
Mail list logo
