On Jun 6, 2010, at 11:00 PM, David Holmes wrote:
cowwoc said the following on 06/07/10 15:25:
That being said, I don't think the PROCESSOR_IDENTIFIER error is
specific
to Make. I noticed the makefile contains echo $
(PROCESSOR_IDENTIFIER).
Running this under cygwin's bash gives:
echo $(PROCESSOR_IDENTIFIER)
bash: PROCESSOR_IDENTIFIER: command not found
while:
echo $PROCESSOR_IDENTIFIER
Intel64 Family 6 Model 15 Stepping 11, GenuineIntel
Where can I get make 3.10 that will work? How do I fix the
PROCESSOR_IDENTIFIER error?
Shouldn't jdk_generic_profile.sh use braces instead of round
brackets? That
is, ${PROCESSOR_IDENTIFIER} instead of $(PROCESSOR_IDENTIFIER)? Is
this a
bug in the makefile?
It's a bug in the script - a typo introduced by this recent changeset:
changeset: 970:90873391a0e0
parent: 945:fea0898259ae
user: ohair
date: Thu Mar 26 16:52:00 2009 -0700
summary: 6822374: Windows: detect X64 when PROCESSOR_IDENTIFIER
contains EM64T or Intel64
diff -r fea0898259ae -r 90873391a0e0 make/jdk_generic_profile.sh
--- a/make/jdk_generic_profile.sh Tue Mar 17 13:45:01 2009 -0700
+++ b/make/jdk_generic_profile.sh Thu Mar 26 16:52:00 2009 -0700
@@ -174,7 +174,8 @@ else
# Check CYGWIN (should have already been done)
# Assumption here is that you are in a shell window via cygwin.
- if [ "$(echo ${PROCESSOR_IDENTIFIER} | fgrep AMD64)" != "" ] ; then
+ proc_arch=`echo "$(PROCESSOR_IDENTIFIER)" | expand | cut -d' ' -
f1 | sed -e 's...@x86@x...@g' -e 's...@intel64@x...@g' -e 's...@em64t@x...@g' -e
's...@em64t@x...@g' -e 's...@amd64@x...@g' -e 's...@amd64@x...@g' -e
's...@ia64@i...@g'`
+ if [ "${proc_arch}" = "X64" ] ; then
windows_arch=amd64
else
windows_arch=i586
My fault. I'll file a bug and fix this. Sorry about that.
I don't think any of our internal automated build processes use this
script, never did.
I added it as more of a form of documentation on the settings needed,
and had assumed people would
copy it and edit to suit their purposes. It was originally created
many years ago, pre-OpenJDK. :^(
If I can figure out a way to somehow test it, I'll try and create some
kind of test. But I'll fix it first.
Anyone care to be the reviewer?
6959116: Fix regression in make/jdk_generic_profile.sh
(PROCESSOR_IDENTIFIER)
diff --git a/make/jdk_generic_profile.sh b/make/jdk_generic_profile.sh
--- a/make/jdk_generic_profile.sh
+++ b/make/jdk_generic_profile.sh
@@ -174,7 +174,7 @@
# Check CYGWIN (should have already been done)
# Assumption here is that you are in a shell window via cygwin.
- proc_arch=`echo "$(PROCESSOR_IDENTIFIER)" | expand | cut -d' ' -f1
| sed -e 's...@x86@x...@g' -e 's...@intel64@x...@g' -e 's...@em64t@x...@g' -e
's...@em64t@x...@g' -e 's...@amd64@x...@g' -e 's...@amd64@x...@g' -e
's...@ia64@i...@g'`
+ proc_arch=`echo "${PROCESSOR_IDENTIFIER}" | expand | cut -d' ' -f1
| sed -e 's...@x86@x...@g' -e 's...@intel64@x...@g' -e 's...@em64t@x...@g' -e
's...@em64t@x...@g' -e 's...@amd64@x...@g' -e 's...@amd64@x...@g' -e
's...@ia64@i...@g'`
if [ "${proc_arch}" = "X64" ] ; then
windows_arch=amd64
else
-kto
David Holmes
-----------
Another thing I learned is that if you get:
"cygpath: cannot create short name of \\?\C:\Program Files\Microsoft
SDKs\Windows\v7.1" it means the referenced path does not exist.
What a
misleading error message :) I initially thought there was something
wrong
with the argument passed to cygpath...
Gili