Re: [OE-core] [PATCH] scripts: Use fixed temporary file instead of pipe for here-doc
On Wed, 2018-11-21 at 18:21 +0800, He Zhe wrote: > > On 2018/11/21 18:07, Richard Purdie wrote: > > On Wed, 2018-11-21 at 17:39 +0800, zhe...@windriver.com wrote: > > > From: He Zhe > > > > > > A workaround for possible "*** Compiler lacks asm-goto support.. > > > Stop." > > > linux-libc-headers is built by gcc on build machine, which could > > > not > > > contain the > > > fix. > > > > > > Signed-off-by: He Zhe > > > --- > > > ...-fixed-temporary-file-instead-of-pipe-for.patch | 60 > > > ++ > > > .../linux-libc-headers/linux-libc-headers_4.18.bb | 4 ++ > > > 2 files changed, 64 insertions(+) > > > create mode 100644 meta/recipes-kernel/linux-libc-headers/linux- > > > libc-headers/0001-scripts-Use-fixed-temporary-file-instead-of- > > > pipe- > > > for.patch > > > > How old is an "old gcc"? > > I put the full explanation in the commit log of the patch inside. No, you did not. I went and looked at the link (which I had to un-line wrap) to understand this and the problem is in binutils in the assembler (as), *not* gcc. Its also not just some "old gcc" but any current released version of binutils which is quite different. Please improve the commit message to reference binutils, not gcc, mention the current version of bintuils (2.31) and that it and all previous versions have the bug. Also please set the Upstream-Status correctly. > The fix below for the bug has not been released from upstream. Since > there has > not been gcc-cross when building linux-libc-headers, we have to use > build > machine gcc which could not contain the fix for the moment. To work > around the > error, we create a fixed temporary file to contain the program being > tested. > > https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;a=commit;h= > 2a50366ded329bfb39d387253450c9d5302c3503 > > > > > If this isn't going upstream, I'm not sure we want to take and > > carry > > this either. Unless this is a gcc people would commonly run into > > I'm > > tempted not to take this. > > This is a gcc issue, but the fix would not quickly goto versions > being used > by users. So this patch is used as a workaround to prevent the > failure. > > To be specific, we encountered this in docker where the /tmp could be > rarely used that very low number inode may be allocated and triggers > the error. That should also be mentioned in the commit message as that does put it in a different light. > > > > Also, your patch subject is wrong as this does not change our > > scripts > > directory. > > If this is OK, I'll send v2 to change the subject. Along with the other changes please. Thanks, Richard -- ___ Openembedded-core mailing list Openembedded-core@lists.openembedded.org http://lists.openembedded.org/mailman/listinfo/openembedded-core
Re: [OE-core] [PATCH] scripts: Use fixed temporary file instead of pipe for here-doc
On 2018/11/21 18:07, Richard Purdie wrote: > On Wed, 2018-11-21 at 17:39 +0800, zhe...@windriver.com wrote: >> From: He Zhe >> >> A workaround for possible "*** Compiler lacks asm-goto support.. >> Stop." >> linux-libc-headers is built by gcc on build machine, which could not >> contain the >> fix. >> >> Signed-off-by: He Zhe >> --- >> ...-fixed-temporary-file-instead-of-pipe-for.patch | 60 >> ++ >> .../linux-libc-headers/linux-libc-headers_4.18.bb | 4 ++ >> 2 files changed, 64 insertions(+) >> create mode 100644 meta/recipes-kernel/linux-libc-headers/linux- >> libc-headers/0001-scripts-Use-fixed-temporary-file-instead-of-pipe- >> for.patch > How old is an "old gcc"? I put the full explanation in the commit log of the patch inside. The fix below for the bug has not been released from upstream. Since there has not been gcc-cross when building linux-libc-headers, we have to use build machine gcc which could not contain the fix for the moment. To work around the error, we create a fixed temporary file to contain the program being tested. https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;a=commit;h= 2a50366ded329bfb39d387253450c9d5302c3503 > > If this isn't going upstream, I'm not sure we want to take and carry > this either. Unless this is a gcc people would commonly run into I'm > tempted not to take this. This is a gcc issue, but the fix would not quickly goto versions being used by users. So this patch is used as a workaround to prevent the failure. To be specific, we encountered this in docker where the /tmp could be rarely used that very low number inode may be allocated and triggers the error. > > Also, your patch subject is wrong as this does not change our scripts > directory. If this is OK, I'll send v2 to change the subject. Thanks, Zhe > > Cheers, > > Richard > > > > -- ___ Openembedded-core mailing list Openembedded-core@lists.openembedded.org http://lists.openembedded.org/mailman/listinfo/openembedded-core
Re: [OE-core] [PATCH] scripts: Use fixed temporary file instead of pipe for here-doc
On Wed, 2018-11-21 at 17:39 +0800, zhe...@windriver.com wrote: > From: He Zhe > > A workaround for possible "*** Compiler lacks asm-goto support.. > Stop." > linux-libc-headers is built by gcc on build machine, which could not > contain the > fix. > > Signed-off-by: He Zhe > --- > ...-fixed-temporary-file-instead-of-pipe-for.patch | 60 > ++ > .../linux-libc-headers/linux-libc-headers_4.18.bb | 4 ++ > 2 files changed, 64 insertions(+) > create mode 100644 meta/recipes-kernel/linux-libc-headers/linux- > libc-headers/0001-scripts-Use-fixed-temporary-file-instead-of-pipe- > for.patch How old is an "old gcc"? If this isn't going upstream, I'm not sure we want to take and carry this either. Unless this is a gcc people would commonly run into I'm tempted not to take this. Also, your patch subject is wrong as this does not change our scripts directory. Cheers, Richard -- ___ Openembedded-core mailing list Openembedded-core@lists.openembedded.org http://lists.openembedded.org/mailman/listinfo/openembedded-core
[OE-core] [PATCH] scripts: Use fixed temporary file instead of pipe for here-doc
From: He Zhe
A workaround for possible "*** Compiler lacks asm-goto support.. Stop."
linux-libc-headers is built by gcc on build machine, which could not contain the
fix.
Signed-off-by: He Zhe
---
...-fixed-temporary-file-instead-of-pipe-for.patch | 60 ++
.../linux-libc-headers/linux-libc-headers_4.18.bb | 4 ++
2 files changed, 64 insertions(+)
create mode 100644
meta/recipes-kernel/linux-libc-headers/linux-libc-headers/0001-scripts-Use-fixed-temporary-file-instead-of-pipe-for.patch
diff --git
a/meta/recipes-kernel/linux-libc-headers/linux-libc-headers/0001-scripts-Use-fixed-temporary-file-instead-of-pipe-for.patch
b/meta/recipes-kernel/linux-libc-headers/linux-libc-headers/0001-scripts-Use-fixed-temporary-file-instead-of-pipe-for.patch
new file mode 100644
index 000..eca3330
--- /dev/null
+++
b/meta/recipes-kernel/linux-libc-headers/linux-libc-headers/0001-scripts-Use-fixed-temporary-file-instead-of-pipe-for.patch
@@ -0,0 +1,60 @@
+From 3bbea65e11918f8753e8006a2198b999cdb0af58 Mon Sep 17 00:00:00 2001
+From: He Zhe
+Date: Wed, 21 Nov 2018 15:12:43 +0800
+Subject: [PATCH] scripts: Use fixed temporary file instead of pipe for
+ here-doc
+
+There was a bug of gcc that when it checks if the input file and output file
are
+the same one, it would not check if they are on the same block device.
+
+The here-doc usage in this script creates temporary file in /tmp. When we run
in
+an environment where /tmp has rarely been used, the newly created temporary
file
+may have a very low inode number. If the inode number was 6 which is the same
as
+/dev/null, gcc would wrongly think the input file and the output file are the
+same and report the following error.
+
+*** Compiler lacks asm-goto support.. Stop.
+
+The fix below for the bug has not been released from upstream. Since there has
+not been gcc-cross when building linux-libc-headers, we have to use build
+machine gcc which could not contain the fix for the moment. To work around the
+error, we create a fixed temporary file to contain the program being tested.
+
+https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;a=commit;h=
+2a50366ded329bfb39d387253450c9d5302c3503
+
+This patch also removes ">/dev/null 2>&1" so we will have more direct error
+information in case something else wrong happened.
+
+Upstream-Status: Inappropriate [Just a work around for old gcc]
+
+Signed-off-by: He Zhe
+---
+ scripts/gcc-goto.sh | 7 ++-
+ 1 file changed, 6 insertions(+), 1 deletion(-)
+
+diff --git a/scripts/gcc-goto.sh b/scripts/gcc-goto.sh
+index 083c526..0aaf1b4 100755
+--- a/scripts/gcc-goto.sh
b/scripts/gcc-goto.sh
+@@ -3,7 +3,9 @@
+ # Test for gcc 'asm goto' support
+ # Copyright (C) 2010, Jason Baron
+
+-cat << "END" | $@ -x c - -c -o /dev/null >/dev/null 2>&1 && echo "y"
++TMPFILE=`mktemp -p .`
++
++cat << "END" > ${TMPFILE}
+ int main(void)
+ {
+ #if defined(__arm__) || defined(__aarch64__)
+@@ -20,3 +22,6 @@ entry:
+ return 0;
+ }
+ END
++
++$@ -x c ${TMPFILE} -c -o /dev/null && echo "y"
++rm ${TMPFILE}
+--
+2.7.4
+
diff --git a/meta/recipes-kernel/linux-libc-headers/linux-libc-headers_4.18.bb
b/meta/recipes-kernel/linux-libc-headers/linux-libc-headers_4.18.bb
index eb7bee7..00420aa 100644
--- a/meta/recipes-kernel/linux-libc-headers/linux-libc-headers_4.18.bb
+++ b/meta/recipes-kernel/linux-libc-headers/linux-libc-headers_4.18.bb
@@ -9,5 +9,9 @@ SRC_URI_append_libc-musl = "\
file://0001-include-linux-stddef.h-in-swab.h-uapi-header.patch \
"
+SRC_URI_append = "\
+file://0001-scripts-Use-fixed-temporary-file-instead-of-pipe-for.patch \
+"
+
SRC_URI[md5sum] = "bee5fe53ee1c3142b8f0c12c0d3348f9"
SRC_URI[sha256sum] =
"19d8bcf49ef530cd4e364a45b4a22fa70714b70349c8100e7308488e26f1eaf1"
--
2.7.4
--
___
Openembedded-core mailing list
Openembedded-core@lists.openembedded.org
http://lists.openembedded.org/mailman/listinfo/openembedded-core
