include/mk/env_post.mk: enable __ANDROID__ definition for Android build
PATCH 1/1 include/mk/env_post.mk: enable __ANDROID__ definition for Android build
Enable the __ANDROID__ definition by default for Android build, otherwise we should manually enable it by configure command.
--- include/mk/env_post.mk | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
include/mk/env_post.mk +1 -1
@@ -44,7 +44,7 @@ endif ifeq ($(ANDROID),1) # There are many undeclared functions, it's best not to accidentally overlook # them. -CFLAGS += -Werror-implicit-function-declaration +CFLAGS += -Werror-implicit-function-declaration -D__ANDROID__ LDFLAGS += -L$(top_builddir)/lib/android_libpthread LDFLAGS += -L$(top_builddir)/lib/android_librt
Signature · 1 line
2.7.4
Could anyone help me to review this patch? This is useful to cross-compile ltp for Android without specifying '-D__ANDROID__' in configure step.
Thanks!
Zhengwang
------------------------------------------------------------------
发件人:Zhengwang Ruan <[email protected]>
发送时间:2019年4月9日(星期二) 12:05
收件人:ltp <[email protected]>
抄 送:Zhengwang Ruan <[email protected]>
主 题:[PATCH v1] include/mk/env_post.mk: enable __ANDROID__ definition for Android build
Enable the __ANDROID__ definition by default for Android build, otherwise we should manually enable it by configure command.
--- include/mk/env_post.mk | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
include/mk/env_post.mk +1 -1
@@ -44,7 +44,7 @@ endif ifeq ($(ANDROID),1) # There are many undeclared functions, it's best not to accidentally overlook # them. -CFLAGS += -Werror-implicit-function-declaration +CFLAGS += -Werror-implicit-function-declaration -D__ANDROID__ LDFLAGS += -L$(top_builddir)/lib/android_libpthread LDFLAGS += -L$(top_builddir)/lib/android_librt
Signature · 5 lines
2.7.4 -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.linux.it/pipermail/ltp/attachments/20190410/8684835f/attachment-0001.html>
Hi Sandeep, Steve,
Quoted 26 lines
Could anyone help me to review this patch? This is useful to cross-compile ltp for Android without specifying '-D__ANDROID__' in configure step.
Thanks!
Zhengwang------------------------------------------------------------------
发件人:Zhengwang Ruan <[email protected]>
发送时间:2019年4月9日(星期二) 12:05
收件人:ltp <[email protected]>
抄 送:Zhengwang Ruan <[email protected]>
主 题:[PATCH v1]include/mk/env_post.mk: enable__ANDROID__definition for Android buildEnable the
__ANDROID__definition by default for Android build, otherwise we should manually enable it by configure command.Signed-off-by Zhengwang Ruan <[email protected]>--- include/mk/env_post.mk | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)include/mk/env_post.mk +0
-0
include/mk/env_post.mk endif · line 47
Quoted 5 lines
include/mk/env_post.mk +1
-1@@ -44,7 +44,7 @@ endif ifeq ($(ANDROID),1) # There are many undeclared functions, it's best not to accidentally overlook # them. -CFLAGS += -Werror-implicit-function-declaration +CFLAGS += -Werror-implicit-function-declaration -D__ANDROID__
Would adding __ANDROID__ definition help building LTP with distro cross-compiler? (i.e. no NDK, no inside AOSP tree)?
I guess so and therefore, I'd be for this change.
I see some projects added it as well (busybox [1] [2], dnsmasq [3]).
BTW: where is defined for AOSP and NDK?
BTW (@cyril): I guess -Werror-implicit-function-declaration is the code which requires adding <string.h> and <strings.h> (see my fix 4053a2551 ("lib: Add missing <string.h> and <strings.h> includes")), I'd be for either adding it to for all or removing it.
include/mk/env_post.mk endif · line 50
Quoted 2 lines
include/mk/env_post.mk +0
-0LDFLAGS += -L$(top_builddir)/lib/android_libpthread LDFLAGS += -L$(top_builddir)/lib/android_librt
Kind regards,
Petr
[1] https://git.busybox.net/busybox/tree/configs/android_defconfig
[2] https://git.busybox.net/busybox/tree/configs/android_ndk_defconfig
[3] http://thekelleys.org.uk/gitweb/?p=dnsmasq.git;a=blob_plain;f=bld/Android.mk;hb=HEAD
Hi Petr,
-------- Original Message --------
From: Petr Vorel
Sent: Wed, 10 Apr 2019 16:16:15 +0200
To: Sandeep Patil, Steve Muckle
Cc: Ltp, Zhengwang Ruan, Cyril Hrubis
Subject: Re: [LTP] 回复:[PATCH v1] include/mk/env_post.mk : enable __ANDROID__ definition for Android build
Quoted 25 lines
Hi Sandeep, Steve,
Quoted 23 lines
Could anyone help me to review this patch? This is useful to cross-compile ltp for Android without specifying '-D__ANDROID__' in configure step. Thanks!
Zhengwang------------------------------------------------------------------
发件人:Zhengwang Ruan <[email protected]>
发送时间:2019年4月9日(星期二) 12:05
收件人:ltp <[email protected]>
抄 送:Zhengwang Ruan <[email protected]>
主 题:[PATCH v1]include/mk/env_post.mk: enable__ANDROID__definition for Android build Enable the__ANDROID__definition by default for Android build, otherwise we should manually enable it by configure command.Signed-off-by Zhengwang Ruan <[email protected]>--- include/mk/env_post.mk | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)include/mk/env_post.mk +0
-0@@ -44,7 +44,7 @@ endif ifeq ($(ANDROID),1) # There are many undeclared functions, it's best not to accidentally overlook
Quoted 8 lines
include/mk/env_post.mk endif · line 47
Quoted 3 lines
include/mk/env_post.mk +1
-1# them. -CFLAGS += -Werror-implicit-function-declaration +CFLAGS += -Werror-implicit-function-declaration -D__ANDROID__Would adding
__ANDROID__definition help building LTP with distro cross-compiler? (i.e. no NDK, no inside AOSP tree)?
I guess so and therefore, I'd be for this change.
I see some projects added it as well (busybox [1] [2], dnsmasq [3]).
BTW: where is defined for AOSP and NDK?
I found that '-DANDROID' instead of '-D__ANDROID__' was passed while cross-compiling binaries in my AOSP project.
Thanks!
Zhengwang
Quoted 6 lines
BTW (@cyril): I guess -Werror-implicit-function-declaration is the code which requires adding <
string.h> and <strings.h> (see my fix4053a2551("lib: Add missing <string.h> and <strings.h> includes")), I'd be for either adding it to for all or removing it.
Quoted 8 lines
include/mk/env_post.mk endif · line 50
Quoted 2 lines
include/mk/env_post.mk +0
-0LDFLAGS += -L$(top_builddir)/lib/android_libpthread LDFLAGS += -L$(top_builddir)/lib/android_librtKind regards,
Petr[1] https://git.busybox.net/busybox/tree/configs/android_defconfig
[2] https://git.busybox.net/busybox/tree/configs/android_ndk_defconfig
[3] http://thekelleys.org.uk/gitweb/?p=dnsmasq.git;a=blob_plain;f=bld/Android.mk;hb=HEAD
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linux.it/pipermail/ltp/attachments/20190411/ff848e3f/attachment.html>
Quoted 23 lines · Petr Vorel
Hi Sandeep, Steve,
Quoted 20 lines
Could anyone help me to review this patch? This is useful to cross-compile ltp for Android without specifying '-D__ANDROID__' in configure step.
Thanks!
Zhengwang------------------------------------------------------------------
发件人:Zhengwang Ruan <[email protected]>
发送时间:2019年4月9日(星期二) 12:05
收件人:ltp <[email protected]>
抄 送:Zhengwang Ruan <[email protected]>
主 题:[PATCH v1]include/mk/env_post.mk: enable__ANDROID__definition for Android buildEnable the
__ANDROID__definition by default for Android build, otherwise we should manually enable it by configure command.Signed-off-by Zhengwang Ruan <[email protected]>--- include/mk/env_post.mk | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
Quoted 15 lines
include/mk/env_post.mk endif · line 47
Quoted 10 lines
include/mk/env_post.mk +1
-1@@ -44,7 +44,7 @@ endif ifeq ($(ANDROID),1) # There are many undeclared functions, it's best not to accidentally overlook # them. -CFLAGS += -Werror-implicit-function-declaration +CFLAGS += -Werror-implicit-function-declaration -D__ANDROID__Would adding
__ANDROID__definition help building LTP with distro cross-compiler? (i.e. no NDK, no inside AOSP tree)?
I guess so and therefore, I'd be for this change.
I see some projects added it as well (busybox [1] [2], dnsmasq [3]).
BTW: where is defined for AOSP and NDK?
I think I've used __ANDROID__ and __BIONIC__ in the past for android-specific things. Honestly, I haven't tried to build for Android outside of AOSP tree yet. I am happy to test this if you tell me how? (sorry for that.)
Steve will also be back by 4/15 and he would know this more than I do. So, Ideally, if we can wait for him it will be great. In the mean time, I can test..
FWIW, I want to be able to do this as well. I know its a lot of pain to have to checkout 100s of gigabytes of AOSP tree in order to build LTP for Android right now. So, let me know how I can test and compare.
<snip>
- ssp
Petr Vorel
Acked-byHi Sandeep,
thanks a lot for your comments.
...
Quoted 13 lines
Quoted 4 lines
Quoted 4 lines
+++
b/include/mk/env_post.mkinclude/mk/env_post.mk +0
-0@@ -44,7 +44,7 @@ endif ifeq ($(ANDROID),1) # There are many undeclared functions, it's best not to accidentally overlookQuoted 8 lines
include/mk/env_post.mk endif · line 47
Quoted 3 lines
include/mk/env_post.mk +1
-1# them. -CFLAGS += -Werror-implicit-function-declaration +CFLAGS += -Werror-implicit-function-declaration -D__ANDROID__Would adding
__ANDROID__definition help building LTP with distro cross-compiler? (i.e. no NDK, no inside AOSP tree)?
I guess so and therefore, I'd be for this change.
I see some projects added it as well (busybox [1] [2], dnsmasq [3]).
BTW: where is defined for AOSP and NDK?
Quoted 3 lines
I think I've used
__ANDROID__and__BIONIC__in the past for android-specific things. Honestly, I haven't tried to build for Android outside of AOSP tree yet. I am happy to test this if you tell me how? (sorry for that.)
Great, thanks a lot!
According to [1] android toolchain (I guess both AOSP and NDK toolchain) define __ANDROID__. Which is what I'd expect.
Grepping aosp only dnsmasq and swiftshader use -D__ANDROID__ (these are from externals). -DANDROID is use more times (e.g. some drivers in hardware - ril, qcom, libsensors and wpa_supplicant).
From my point of view, instead of -D__ANDROID__ or -DANDROID I'd prefer to to have proper #ifdef HAVE_FOO guarders made by autotools checks. As android is not the only minor libc. But I understand it's not always possible and we already use some #ifdef __ANDROID__ in library sources and headers, so ack from myself to help compilation outside AOSP and without NDK.
Quoted 3 lines
Steve will also be back by 4/15 and he would know this more than I do. So, Ideally, if we can wait for him it will be great. In the mean time, I can test..
Sure, this can wait for him.
Quoted 3 lines
FWIW, I want to be able to do this as well. I know its a lot of pain to have to checkout 100s of gigabytes of AOSP tree in order to build LTP for Android right now. So, let me know how I can test and compare.
Simple 'make ANDROID=1' will do the build ANDROID is for skipping some tests which are not android ready (+ this patch). I guess without this you probably also want to remove -Werror-implicit-function-declaration (I haven't pushed all my fixes of missing <string.h> and <strings.h> as this is infinite work so far.
I have in my private todo list to write some doc about building LTP for Android. But probably you, Android devs, are more appropriate people :).
Quoted 1 line
- ssp
Kind regards,
Petr
[1] https://groups.google.com/forum/#!topic/android-ndk/cf9_f1SLXls
Hi Sandeep,
Quoted 20 lines
Quoted 11 lines
Quoted 7 lines
include/mk/env_post.mk endif · line 47
Quoted 2 lines
include/mk/env_post.mk +1
-1-CFLAGS += -Werror-implicit-function-declaration +CFLAGS += -Werror-implicit-function-declaration -D__ANDROID__Would adding
__ANDROID__definition help building LTP with distro cross-compiler? (i.e. no NDK, no inside AOSP tree)?
I guess so and therefore, I'd be for this change.
I see some projects added it as well (busybox [1] [2], dnsmasq [3]).
BTW: where is defined for AOSP and NDK?I think I've used
__ANDROID__and__BIONIC__in the past for android-specific things. Honestly, I haven't tried to build for Android outside of AOSP tree yet. I am happy to test this if you tell me how? (sorry for that.)Great, thanks a lot!
According to [1] android toolchain (I guess both AOSP and NDK toolchain) define
__ANDROID__. Which is what I'd expect.Grepping aosp only dnsmasq and swiftshader use -D__ANDROID__ (these are from externals). -DANDROID is use more times (e.g. some drivers in hardware - ril, qcom, libsensors and
wpa_supplicant).
Quoted 5 lines
From my point of view, instead of -D__ANDROID__ or -DANDROID I'd prefer to to have proper #ifdef
HAVE_FOOguarders made by autotools checks. As android is not the only minor libc. But I understand it's not always possible and we already use some #ifdef__ANDROID__in library sources and headers, so ack from myself to help compilation outside AOSP and without NDK.
Actually I was wrong. 1) I was using NDK toolchain. I'd expect it'd generate __ANDROID__, but I don't see it in the output. 2) Building for android with distro toolchain doesn't make sense as it wouldn't have Bionic.
Kind regards,
Petr
Hi Petr,
-------- Original Message --------
From: Petr Vorel
Sent: Fri, 12 Apr 2019 01:40:42 +0200
To: Sandeep Patil
Cc: Ltp
Subject: Re: [LTP] 回复:[PATCH v1] include/mk/env_post.mk: enable __ANDROID__ definition for Android b uild
Quoted 23 lines
Hi Sandeep,
Quoted 10 lines
Quoted 10 lines
Quoted 7 lines
include/mk/env_post.mk endif · line 47
Quoted 2 lines
include/mk/env_post.mk +1
-1-CFLAGS += -Werror-implicit-function-declaration +CFLAGS += -Werror-implicit-function-declaration -D__ANDROID__Would adding
__ANDROID__definition help building LTP with distro cross-compiler? (i.e. no NDK, no inside AOSP tree)?
I guess so and therefore, I'd be for this change.
I see some projects added it as well (busybox [1] [2], dnsmasq [3]).
BTW: where is defined for AOSP and NDK?I think I've used
__ANDROID__and__BIONIC__in the past for android-specific things. Honestly, I haven't tried to build for Android outside of AOSP tree yet. I am happy to test this if you tell me how? (sorry for that.)Quoted 11 lines
Great, thanks a lot!
According to [1] android toolchain (I guess both AOSP and NDK toolchain) define__ANDROID__. Which is what I'd expect.
Grepping aosp only dnsmasq and swiftshader use -D__ANDROID__ (these are from externals). -DANDROID is use more times (e.g. some drivers in hardware - ril, qcom, libsensors andwpa_supplicant).
From my point of view, instead of -D__ANDROID__ or -DANDROID I'd prefer to to have
proper #ifdefHAVE_FOOguarders made by autotools checks. As android is not the only minor libc. But I understand it's not always possible and we already use some #ifdef__ANDROID__in library sources and headers, so ack from myself to help compilation outside AOSP and without NDK.
Quoted 3 lines
Actually I was wrong. 1) I was using NDK toolchain. I'd expect it'd generate
__ANDROID__, but I don't see it in the output. 2) Building for android with distro toolchain doesn't make sense as it wouldn't have Bionic.
1) Did you build outside AOSP? If so, I think it is reasonable since whatever __ANDROID__ or ANDROID are defined by Android build system instead of toolchain. I also tried to build the external/* thing and found ANDROID was actually generated. In my test, ANDROID seems to be the more suitable one, but in fact we can not build ltp inside AOSP because there is no Android.mk in ltp (please correct me if I am wrong...), so ANDROID is meaningless to us here. Actually, this is why I think this patch is still useful, since __ANDROID__ is enough to distinguish what to compile if ANDROID=1 is set with make.
2) In the case of cross-compilation, I think it still makes sense, since we can use the same toolchain, such as clang, and build parameters (the specify the include headers and libs, etc) as AOSP. This is what I am doing now, you can see my steps I posted in my previous reply. But in this case, we need to obviously set -D__ANDROID__ in CFLAGS. After that , the process should be similar as we launch a build inside AOSP.
Regards,
Zhengwang
Quoted 4 lines
Kind regards,
Petr
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linux.it/pipermail/ltp/attachments/20190412/ce8bc90a/attachment.html>
Hi Zhengwang,
Quoted 10 lines
Quoted 10 lines
Quoted 10 lines
According to [1] android toolchain (I guess both AOSP and NDK toolchain) define
__ANDROID__. Which is what I'd expect.
Grepping aosp only dnsmasq and swiftshader use -D__ANDROID__ (these are from externals). -DANDROID is use more times (e.g. some drivers in hardware - ril, qcom, libsensors andwpa_supplicant).
From my point of view, instead of -D__ANDROID__ or -DANDROID I'd prefer to to have
proper #ifdefHAVE_FOOguarders made by autotools checks. As android is not the only minor libc. But I understand it's not always possible and we already use some #ifdef__ANDROID__in library sources and headers, so ack from myself to help compilation outside AOSP and without NDK.
Quoted 5 lines
Quoted 3 lines
Actually I was wrong. 1) I was using NDK toolchain. I'd expect it'd generate
__ANDROID__, but I don't see it in the output. 2) Building for android with distro toolchain doesn't make sense as it wouldn't have Bionic.1) Did you build outside AOSP? If so, I think it is reasonable since
Yes, with android NDK (for Bionic and special toolchain tweeks). Then I use ANDROID=1
Quoted 5 lines
whatever
__ANDROID__or ANDROID are defined by Android build system instead of toolchain. I also tried to build the external/* thing and found ANDROID was actually generated. In my test, ANDROID seems to be the more suitable one, but in fact we can not build ltp inside AOSP because there is no Android.mk in ltp (please correct me if I am wrong...), so ANDROID is
Yes, inside AOSP you need to use Android fork [1] which uses usual Android.bp. Therefore AOSP does not use our build system.
Quoted 10 lines
meaningless to us here. Actually, this is why I think this patch is still useful, since
__ANDROID__is enough to distinguish what to compile if ANDROID=1 is set with make.2) In the case of cross-compilation, I think it still makes sense, since we can use the same toolchain, such as clang, and build parameters (the specify the include headers and libs, etc) as AOSP. This is what I am doing now, you can see my steps I posted in my previous reply. But in this case, we need to obviously set -D__ANDROID__ in CFLAGS. After that , the process should be similar as we launch a build inside AOSP.
I don't know much about android toolchain, but IMHO there are some other tweaks in their build system (to comply changes in Bionic), so not sure if it's even possible to avoid using NDK for out-of-aosp tree. See also [2]. But that's a question for for Android devs.
Quoted 3 lines
Regards,
Zhengwang
Kind regards,
Petr
[1] https://android.googlesource.com/platform/external/ltp/
[2] https://groups.google.com/forum/#!topic/android-ndk/cf9_f1SLXls
-------- Original Message --------
From: Petr Vorel
Sent: Fri, 12 Apr 2019 08:28:14 +0200
To: 阮正旺
Cc: Sandeep Patil, Ltp, Steve Muckle
Subject: Re: [LTP] 回复:[PATCH v1] include/mk/env_post.mk : enable __ANDROID__ definition for Android build
Quoted 36 lines
Hi Zhengwang,
Quoted 10 lines
Quoted 10 lines
Quoted 10 lines
According to [1] android toolchain (I guess both AOSP and NDK toolchain) define
__ANDROID__. Which is what I'd expect.
Grepping aosp only dnsmasq and swiftshader use -D__ANDROID__ (these are from externals). -DANDROID is use more times (e.g. some drivers in hardware - ril, qcom, libsensors andwpa_supplicant).
From my point of view, instead of -D__ANDROID__ or -DANDROID I'd prefer to to have
proper #ifdefHAVE_FOOguarders made by autotools checks. As android is not the only minor libc. But I understand it's not always possible and we already use some #ifdef__ANDROID__in library sources and headers, so ack from myself to help compilation outside AOSP and without NDK.Quoted 4 lines
Quoted 3 lines
Actually I was wrong. 1) I was using NDK toolchain. I'd expect it'd generate
__ANDROID__, but I don't see it in the output. 2) Building for android with distro toolchain doesn't make sense as it wouldn't have Bionic.1) Did you build outside AOSP? If so, I think it is reasonable since
Yes, with android NDK (for Bionic and special toolchain tweeks). Then I use ANDROID=1
Quoted 5 lines
whatever
__ANDROID__or ANDROID are defined by Android build system instead of toolchain. I also tried to build the external/* thing and found ANDROID was actually generated. In my test, ANDROID seems to be the more suitable one, but in fact we can not build ltp inside AOSP because there is no Android.mk in ltp (please correct me if I am wrong...), so ANDROID isYes, inside AOSP you need to use Android fork [1] which uses usual Android.bp. Therefore AOSP does not use our build system.
Quoted 9 lines
meaningless to us here. Actually, this is why I think this patch is still useful, since
__ANDROID__is enough to distinguish what to compile if ANDROID=1 is set with make.
2) In the case of cross-compilation, I think it still makes sense, since we can use the same toolchain, such as clang, and build parameters (the specify the include headers and libs, etc) as AOSP. This is what I am doing now, you can see my steps I posted in my previous reply. But in this case, we need to obviously set -D__ANDROID__ in CFLAGS. After that , the process should be similar as we launch a build inside AOSP.
Quoted 4 lines
I don't know much about android toolchain, but IMHO there are some other tweaks in their build system (to comply changes in Bionic), so not sure if it's even possible to avoid using NDK for out-of-aosp tree. See also [2]. But that's a question for for Android devs.
According to also [2], the situation gets clear now. All android targets should get __ANDROID__ defined by the compiler, but ANDROID only means 'this is AOSP' and is only for Windows targets. In our case, __ANDROID__ should be preferred.
Regards,
Zhengwang
Quoted 8 lines
Quoted 3 lines
Regards,
ZhengwangKind regards,
Petr[1] https://android.googlesource.com/platform/external/ltp/
[2] https://groups.google.com/forum/#!topic/android-ndk/cf9_f1SLXls
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linux.it/pipermail/ltp/attachments/20190413/747d6cf9/attachment.html>
Quoted 49 lines · 阮正旺
-------- Original Message --------
From: Petr Vorel
Sent: Fri, 12 Apr 2019 08:28:14 +0200
To: 阮正旺
Cc: Sandeep Patil, Ltp, Steve Muckle
Subject: Re: [LTP] 回复:[PATCH v1]include/mk/env_post.mk: enable__ANDROID__definition for Android buildQuoted 36 lines
Hi Zhengwang,
Quoted 10 lines
Quoted 10 lines
Quoted 10 lines
According to [1] android toolchain (I guess both AOSP and NDK toolchain) define
__ANDROID__. Which is what I'd expect.
Grepping aosp only dnsmasq and swiftshader use -D__ANDROID__ (these are from externals). -DANDROID is use more times (e.g. some drivers in hardware - ril, qcom, libsensors andwpa_supplicant).
From my point of view, instead of -D__ANDROID__ or -DANDROID I'd prefer to to have
proper #ifdefHAVE_FOOguarders made by autotools checks. As android is not the only minor libc. But I understand it's not always possible and we already use some #ifdef__ANDROID__in library sources and headers, so ack from myself to help compilation outside AOSP and without NDK.Quoted 4 lines
Quoted 3 lines
Actually I was wrong. 1) I was using NDK toolchain. I'd expect it'd generate
__ANDROID__, but I don't see it in the output. 2) Building for android with distro toolchain doesn't make sense as it wouldn't have Bionic.1) Did you build outside AOSP? If so, I think it is reasonable since
Yes, with android NDK (for Bionic and special toolchain tweeks). Then I use ANDROID=1
Quoted 5 lines
whatever
__ANDROID__or ANDROID are defined by Android build system instead of toolchain. I also tried to build the external/* thing and found ANDROID was actually generated. In my test, ANDROID seems to be the more suitable one, but in fact we can not build ltp inside AOSP because there is no Android.mk in ltp (please correct me if I am wrong...), so ANDROID isYes, inside AOSP you need to use Android fork [1] which uses usual Android.bp. Therefore AOSP does not use our build system.
Quoted 9 lines
meaningless to us here. Actually, this is why I think this patch is still useful, since
__ANDROID__is enough to distinguish what to compile if ANDROID=1 is set with make.
2) In the case of cross-compilation, I think it still makes sense, since we can use the same toolchain, such as clang, and build parameters (the specify the include headers and libs, etc) as AOSP. This is what I am doing now, you can see my steps I posted in my previous reply. But in this case, we need to obviously set -D__ANDROID__ in CFLAGS. After that , the process should be similar as we launch a build inside AOSP.Quoted 4 lines
I don't know much about android toolchain, but IMHO there are some other tweaks in their build system (to comply changes in Bionic), so not sure if it's even possible to avoid using NDK for out-of-aosp tree. See also [2]. But that's a question for for Android devs.
Quoted 4 lines
According to also [2], the situation gets clear now. All android targets should get
__ANDROID__defined by the compiler, but ANDROID only means 'this is AOSP' and is only for Windows targets. In our case,__ANDROID__should be preferred.
Correct, and the build can still be make ANDROID=1 which translates to -D__ANDROID__.
JFYI, there have been occasions when we changed something in bionic as a result of an LTP tests, so those fixes will be missing for someone who is using ndk for testing and building LTP.
That being said, thats a fairly small subset and is manageable. We can catch those in reviews / patches very easily.
I like the idea of being able to build ouf-of-aosp, so I'm going to give this a try now and will report.
- ssp
Quoted 14 lines
Regards,
Zhengwang
Quoted 8 lines
Quoted 3 lines
Regards,
ZhengwangKind regards,
Petr[1] https://android.googlesource.com/platform/external/ltp/
[2] https://groups.google.com/forum/#!topic/android-ndk/cf9_f1SLXls
Hi Sandeep,
-------- Original Message --------
From: Sandeep Patil
Sent: Fri, 12 Apr 2019 12:48:09 -0700
To: 阮正旺
Cc: Petr Vorel, Ltp, Steve Muckle
Subject: Re: [LTP] 回复:[PATCH v1] include/mk/env_post.mk : enable __ANDROID__ definition for Android build
Quoted 59 lines · 阮正旺
Quoted 43 lines
-------- Original Message --------
From: Petr Vorel
Sent: Fri, 12 Apr 2019 08:28:14 +0200
To: 阮正旺
Cc: Sandeep Patil, Ltp, Steve Muckle
Subject: Re: [LTP] 回复:[PATCH v1]include/mk/env_post.mk: enable__ANDROID__definition for Android buildQuoted 36 lines
Hi Zhengwang,
Quoted 10 lines
Quoted 10 lines
Quoted 10 lines
According to [1] android toolchain (I guess both AOSP and NDK toolchain) define
__ANDROID__. Which is what I'd expect.
Grepping aosp only dnsmasq and swiftshader use -D__ANDROID__ (these are from externals). -DANDROID is use more times (e.g. some drivers in hardware - ril, qcom, libsensors andwpa_supplicant).
From my point of view, instead of -D__ANDROID__ or -DANDROID I'd prefer to to have
proper #ifdefHAVE_FOOguarders made by autotools checks. As android is not the only minor libc. But I understand it's not always possible and we already use some #ifdef__ANDROID__in library sources and headers, so ack from myself to help compilation outside AOSP and without NDK.Quoted 4 lines
Quoted 3 lines
Actually I was wrong. 1) I was using NDK toolchain. I'd expect it'd generate
__ANDROID__, but I don't see it in the output. 2) Building for android with distro toolchain doesn't make sense as it wouldn't have Bionic.1) Did you build outside AOSP? If so, I think it is reasonable since
Yes, with android NDK (for Bionic and special toolchain tweeks). Then I use ANDROID=1
Quoted 5 lines
whatever
__ANDROID__or ANDROID are defined by Android build system instead of toolchain. I also tried to build the external/* thing and found ANDROID was actually generated. In my test, ANDROID seems to be the more suitable one, but in fact we can not build ltp inside AOSP because there is no Android.mk in ltp (please correct me if I am wrong...), so ANDROID isYes, inside AOSP you need to use Android fork [1] which uses usual Android.bp. Therefore AOSP does not use our build system.
Quoted 9 lines
meaningless to us here. Actually, this is why I think this patch is still useful, since
__ANDROID__is enough to distinguish what to compile if ANDROID=1 is set with make.
2) In the case of cross-compilation, I think it still makes sense, since we can use the same toolchain, such as clang, and build parameters (the specify the include headers and libs, etc) as AOSP. This is what I am doing now, you can see my steps I posted in my previous reply. But in this case, we need to obviously set -D__ANDROID__ in CFLAGS. After that , the process should be similar as we launch a build inside AOSP.Quoted 8 lines
Quoted 4 lines
I don't know much about android toolchain, but IMHO there are some other tweaks in their build system (to comply changes in Bionic), so not sure if it's even possible to avoid using NDK for out-of-aosp tree. See also [2]. But that's a question for for Android devs.
According to also [2], the situation gets clear now. All android targets should get
__ANDROID__defined by the compiler, but ANDROID only means 'this is AOSP' and is only for Windows targets. In our case,__ANDROID__should be preferred.Correct, and the build can still be make ANDROID=1 which translates to -D__ANDROID__.
JFYI, there have been occasions when we changed something in bionic as a result of an LTP tests, so those fixes will be missing for someone who is using ndk for testing and building LTP.
Quoted 5 lines
That being said, thats a fairly small subset and is manageable. We can catch those in reviews / patches very easily.
I like the idea of being able to build ouf-of-aosp, so I'm going to give this a try now and will report.
Have you tried to build out-of-aosp? :-)
Regards,
Zhengwang
Quoted 13 lines
- ssp
Quoted 12 lines
Regards,
Zhengwang
Quoted 7 lines
Quoted 2 lines
Regards,
ZhengwangKind regards,
Petr[1] https://android.googlesource.com/platform/external/ltp/
[2] https://groups.google.com/forum/#!topic/android-ndk/cf9_f1SLXls
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linux.it/pipermail/ltp/attachments/20190416/cb4979bb/attachment.html>
Quoted 69 lines · 阮正旺
Hi Sandeep,
-------- Original Message --------
From: Sandeep Patil
Sent: Fri, 12 Apr 2019 12:48:09 -0700
To: 阮正旺
Cc: Petr Vorel, Ltp, Steve Muckle
Subject: Re: [LTP] 回复:[PATCH v1]include/mk/env_post.mk: enable__ANDROID__definition for Android buildQuoted 59 lines · 阮正旺
Quoted 43 lines
-------- Original Message --------
From: Petr Vorel
Sent: Fri, 12 Apr 2019 08:28:14 +0200
To: 阮正旺
Cc: Sandeep Patil, Ltp, Steve Muckle
Subject: Re: [LTP] 回复:[PATCH v1]include/mk/env_post.mk: enable__ANDROID__definition for Android buildQuoted 36 lines
Hi Zhengwang,
Quoted 10 lines
Quoted 10 lines
Quoted 10 lines
According to [1] android toolchain (I guess both AOSP and NDK toolchain) define
__ANDROID__. Which is what I'd expect.
Grepping aosp only dnsmasq and swiftshader use -D__ANDROID__ (these are from externals). -DANDROID is use more times (e.g. some drivers in hardware - ril, qcom, libsensors andwpa_supplicant).
From my point of view, instead of -D__ANDROID__ or -DANDROID I'd prefer to to have
proper #ifdefHAVE_FOOguarders made by autotools checks. As android is not the only minor libc. But I understand it's not always possible and we already use some #ifdef__ANDROID__in library sources and headers, so ack from myself to help compilation outside AOSP and without NDK.Quoted 4 lines
Quoted 3 lines
Actually I was wrong. 1) I was using NDK toolchain. I'd expect it'd generate
__ANDROID__, but I don't see it in the output. 2) Building for android with distro toolchain doesn't make sense as it wouldn't have Bionic.1) Did you build outside AOSP? If so, I think it is reasonable since
Yes, with android NDK (for Bionic and special toolchain tweeks). Then I use ANDROID=1
Quoted 5 lines
whatever
__ANDROID__or ANDROID are defined by Android build system instead of toolchain. I also tried to build the external/* thing and found ANDROID was actually generated. In my test, ANDROID seems to be the more suitable one, but in fact we can not build ltp inside AOSP because there is no Android.mk in ltp (please correct me if I am wrong...), so ANDROID isYes, inside AOSP you need to use Android fork [1] which uses usual Android.bp. Therefore AOSP does not use our build system.
Quoted 9 lines
meaningless to us here. Actually, this is why I think this patch is still useful, since
__ANDROID__is enough to distinguish what to compile if ANDROID=1 is set with make.
2) In the case of cross-compilation, I think it still makes sense, since we can use the same toolchain, such as clang, and build parameters (the specify the include headers and libs, etc) as AOSP. This is what I am doing now, you can see my steps I posted in my previous reply. But in this case, we need to obviously set -D__ANDROID__ in CFLAGS. After that , the process should be similar as we launch a build inside AOSP.Quoted 8 lines
Quoted 4 lines
I don't know much about android toolchain, but IMHO there are some other tweaks in their build system (to comply changes in Bionic), so not sure if it's even possible to avoid using NDK for out-of-aosp tree. See also [2]. But that's a question for for Android devs.
According to also [2], the situation gets clear now. All android targets should get
__ANDROID__defined by the compiler, but ANDROID only means 'this is AOSP' and is only for Windows targets. In our case,__ANDROID__should be preferred.Correct, and the build can still be make ANDROID=1 which translates to -D__ANDROID__.
JFYI, there have been occasions when we changed something in bionic as a result of an LTP tests, so those fixes will be missing for someone who is using ndk for testing and building LTP.
Quoted 7 lines
Quoted 5 lines
That being said, thats a fairly small subset and is manageable. We can catch those in reviews / patches very easily.
I like the idea of being able to build ouf-of-aosp, so I'm going to give this a try now and will report.
Have you tried to build out-of-aosp? :-)
Yes, I did that now and ran into a build error with pread02 pasted at the end of this email.
This was much easier than I thought and I think I can start sending trivial fixes as I go. Here's what I did.
1. Download the ndk20 beta from - https://developer.android.com/ndk/downloads
2. Set up ndk toolchains for autoconf as shown in
https://developer.android.com/ndk/guides/other_build_systems#autoconf
3. In LTP, do
$ ./configure --host x86_64-linux-android --prefix=<preferred install location> $ make ANDROID=1
Please let me know if this doesn't sound right? It seems a lot easier than the steps pasted earlier.
- ssp
======================
/usr/local/google/home/sspatil/bin/android-ndk-r20-beta2/toolchains/llvm/prebuilt/linux-x86_64/bin/x86_64-linux-android28-clang -g -O2 -g -O2 -fno-strict-aliasing -pipe -Wall -W -Wold-style-definition -Werror-implicit-function-declaration -D_FORTIFY_SOURCE=2 -I/usr/local/google/home/sspatil/dev/ssd2/ltp/testcases/kernel/syscalls/preadv -I/usr/local/google/home/sspatil/dev/ssd2/ltp/testcases/kernel/syscalls/preadv/../utils -I../../../../include -I../../../../include -I../../../../include/old/ -L../../../../lib -L../../../../lib/android_libpthread -L../../../../lib/android_librt preadv02.c -lltp -o preadv02
preadv02.c:89:7: error: implicit declaration of function 'preadv' is invalid in C99 [-Werror,-Wimplicit-function-declaration]
TEST(preadv(*tc->fd, tc->name, tc->count, tc->offset));
^1 error generated.
<builtin>: recipe for target 'preadv02' failed
make[4]: *** [preadv02] Error 1
make[4]: Leaving directory '/usr/local/google/home/sspatil/dev/ssd2/ltp/testcases/kernel/syscalls/preadv' ../../../include/mk/generic_trunk_target.inc:93: recipe for target 'all' failed
make[3]: *** [all] Error 2
make[3]: Leaving directory '/usr/local/google/home/sspatil/dev/ssd2/ltp/testcases/kernel/syscalls' ../../include/mk/generic_trunk_target.inc:93: recipe for target 'all' failed
make[2]: *** [all] Error 2
make[2]: Leaving directory '/usr/local/google/home/sspatil/dev/ssd2/ltp/testcases/kernel' ../include/mk/generic_trunk_target.inc:93: recipe for target 'all' failed
make[1]: *** [all] Error 2
make[1]: Leaving directory '/usr/local/google/home/sspatil/dev/ssd2/ltp/testcases' Makefile:111: recipe for target 'testcases-all' failed
make: *** [testcases-all] Error 2
Hi Sandeep,
Quoted 11 lines
Quoted 1 line
Have you tried to build out-of-aosp? :-)
Yes, I did that now and ran into a build error with pread02 pasted at the end of this email.
This was much easier than I thought and I think I can start sending trivial fixes as I go. Here's what I did.
1. Download the ndk20 beta from - https://developer.android.com/ndk/downloads
2. Set up ndk toolchains for autoconf as shown in
https://developer.android.com/ndk/guides/other_build_systems#autoconf
Thanks for a hint. I was using standalone toolchain [1], created with NKD r19, but see according a doc that it was deprecated even for r19.
[1] https://developer.android.com/ndk/guides/standalone_toolchain
Quoted 3 lines
3. In LTP, do
$ ./configure --host x86_64-linux-android --prefix=<preferred install location> $ make ANDROID=1
I'm also passing these flags to CFLAGS, but it might not be necessary:
--sysroot=${SYSROOT} -I${SYSROOT}/usr/include -I${ANDROID_PREFIX}/include
for arm host I use:
--host arm-linux-androideabi
and also define CROSS_COMPILE + other variables:
export CROSS_COMPILE=$TOOLCHAIN_ROOT/bin/arm-linux-androideabi- export AR=${CROSS_COMPILE}ar
export AS=${CROSS_COMPILE}as
export NM=${CROSS_COMPILE}nm
export CC=${CROSS_COMPILE}gcc
export LD=${CROSS_COMPILE}ld
export RANLIB=${CROSS_COMPILE}ranlib
Only CROSS_COMPILE should be needed as, others are in include/mk/config.mk.default, but my toolchain setup does not detect them (include/mk/config.mk does not have cross-compile versions)
Quoted 6 lines
Please let me know if this doesn't sound right? It seems a lot easier than the steps pasted earlier.
- ssp
Quoted 6 lines
======================
/usr/local/google/home/sspatil/bin/android-ndk-r20-beta2/toolchains/llvm/prebuilt/linux-x86_64/bin/x86_64-linux-android28-clang-g -O2 -g -O2 -fno-strict-aliasing -pipe -Wall -W -Wold-style-definition -Werror-implicit-function-declaration -D_FORTIFY_SOURCE=2 -I/usr/local/google/home/sspatil/dev/ssd2/ltp/testcases/kernel/syscalls/preadv -I/usr/local/google/home/sspatil/dev/ssd2/ltp/testcases/kernel/syscalls/preadv/../utils -I../../../../include -I../../../../include -I../../../../include/old/ -L../../../../lib -L../../../../lib/android_libpthread -L../../../../lib/android_librtpreadv02.c-lltp -o preadv02
preadv02.c:89:7: error: implicit declaration of function 'preadv' is invalid in C99 [-Werror,-Wimplicit-function-declaration]TEST(preadv(*tc->fd, tc->name, tc->count, tc->offset)); ^
I bump into the same problem
+ this one:
pwritev02.c:85:7: error: implicit declaration of function 'pwritev' is invalid in C99 [-Werror,-Wimplicit-function-declaration]
TEST(pwritev(*tc->fd, tc->name, tc->count, tc->offset));
So for the build is needed to comment out HAVE_PREADV and HAVE_PWRITEV definitions in include/config.h.
I don't know why autotools fails preadv and pwritev detection with AC_CHECK_FUNCS. I considered it as problem of my setup, but maybe it's an autotools bug.
Quoted 7 lines
1 error generated.
<builtin>: recipe for target 'preadv02' failed
make[4]: *** [preadv02] Error 1
make[4]: Leaving directory '/usr/local/google/home/sspatil/dev/ssd2/ltp/testcases/kernel/syscalls/preadv' ../../../include/mk/generic_trunk_target.inc:93: recipe for target 'all' failed
make[3]: *** [all] Error 2
make[3]: Leaving directory '/usr/local/google/home/sspatil/dev/ssd2/ltp/testcases/kernel/syscalls'
Quoted 8 lines
../../include/mk/generic_trunk_target.inc:93: recipe for target 'all' failed make[2]: *** [all] Error 2
make[2]: Leaving directory '/usr/local/google/home/sspatil/dev/ssd2/ltp/testcases/kernel' ../include/mk/generic_trunk_target.inc:93: recipe for target 'all' failed make[1]: *** [all] Error 2
make[1]: Leaving directory '/usr/local/google/home/sspatil/dev/ssd2/ltp/testcases' Makefile:111: recipe for target 'testcases-all' failed
make: *** [testcases-all] Error 2
Kind regards,
Petr
Hi Sandeep, Petr,
-------- Original Message --------
From: Petr Vorel
Sent: Wed, 17 Apr 2019 19:06:51 +0200
To: Sandeep Patil
Cc: Zhengwang Ruan, Ltp, Steve Muckle
Subject: Re: [LTP] 回复:[PATCH v1] include/mk/env_post.mk : enable __ANDROID__ definition for Android build
Quoted 50 lines
Hi Sandeep,
Quoted 8 lines
Quoted 1 line
Have you tried to build out-of-aosp? :-)
Yes, I did that now and ran into a build error with pread02 pasted at the end of this email.
This was much easier than I thought and I think I can start sending trivial fixes as I go. Here's what I did.
1. Download the ndk20 beta from - https://developer.android.com/ndk/downloads
2. Set up ndk toolchains for autoconf as shown in
https://developer.android.com/ndk/guides/other_build_systems#autoconfThanks for a hint. I was using standalone toolchain [1], created with NKD r19, but see according a doc that it was deprecated even for r19.
[1] https://developer.android.com/ndk/guides/standalone_toolchain
Quoted 3 lines
3. In LTP, do
$ ./configure --host x86_64-linux-android --prefix=<preferred install location> $ make ANDROID=1I'm also passing these flags to CFLAGS, but it might not be necessary:
--sysroot=${SYSROOT} -I${SYSROOT}/usr/include -I${ANDROID_PREFIX}/includefor arm host I use:
--host arm-linux-androideabiand also define
CROSS_COMPILE+ other variables:
export CROSS_COMPILE=$TOOLCHAIN_ROOT/bin/arm-linux-androideabi- export AR=${CROSS_COMPILE}ar
export AS=${CROSS_COMPILE}as
export NM=${CROSS_COMPILE}nm
export CC=${CROSS_COMPILE}gcc
export LD=${CROSS_COMPILE}ld
export RANLIB=${CROSS_COMPILE}ranlibOnly
CROSS_COMPILEshould be needed as, others are ininclude/mk/config.mk.default, but my toolchain setup does not detect them (include/mk/config.mkdoes not have cross-compile versions)Quoted 9 lines
Please let me know if this doesn't sound right? It seems a lot easier than the steps pasted earlier.
- ssp======================
/usr/local/google/home/sspatil/bin/android-ndk-r20-beta2/toolchains/llvm/prebuilt/linux-x86_64/bin/x86_64-linux-android28-clang-g -O2 -g -O2 -fno-strict-aliasing -pipe -Wall -W -Wold-style-definition -Werror-implicit-function-declaration -D_FORTIFY_SOURCE=2 -I/usr/local/google/home/sspatil/dev/ssd2/ltp/testcases/kernel/syscalls/preadv -I/usr/local/google/home/sspatil/dev/ssd2/ltp/testcases/kernel/syscalls/preadv/../utils -I../../../../include -I../../../../include -I../../../../include/old/ -L../../../../lib -L../../../../lib/android_libpthread -L../../../../lib/android_librtpreadv02.c-lltp -o preadv02
preadv02.c:89:7: error: implicit declaration of function 'preadv' is invalid in C99 [-Werror,-Wimplicit-function-declaration]TEST(preadv(*tc->fd, tc->name, tc->count, tc->offset)); ^I bump into the same problem
+ this one:
pwritev02.c:85:7: error: implicit declaration of function 'pwritev' is invalid in C99 [-Werror,-Wimplicit-function-declaration]
Quoted 7 lines
TEST(pwritev(*tc->fd, tc->name, tc->count, tc->offset));So for the build is needed to comment out
HAVE_PREADVandHAVE_PWRITEVdefinitions ininclude/config.h.I don't know why autotools fails preadv and pwritev detection with
AC_CHECK_FUNCS. I considered it as problem of my setup, but maybe it's an autotools bug.
I also encountered this issue, but I temporarily walked around it by applying the following patch.
==================
$ git diff testcases/
(quoted diff) +0 -0
b/testcases/commands/ldd/datafiles/Makefile
index d8bcf5c..74fbe50 100644
testcases/commands/ldd/datafiles/Makefile +1 -1
@@ -30,7 +30,7 @@ CLEAN_TARGETS += *.obj $(LDD_SO_FILES) $(CC) $(CFLAGS) -shared -o $@ $^ lddfile.out: main.o $(LDD_SO_FILES) - $(CC) $(CFLAGS) -O -o $@ $? + $(CC) $(CFLAGS) $(LDFLAGS) -O -o $@ $? INSTALL_DIR := testcases/data/ldd01
(quoted diff) +0 -0
b/testcases/kernel/syscalls/Makefile
index 45a00cf..27c2e7a 100644
testcases/kernel/syscalls/Makefile +0 -0
@@ -43,6 +43,8 @@ FILTER_OUT_DIRS += \ sched_getaffinity sched_setaffinity sendmsg setgroups setns
sighold \
sigrelse sigsuspend sigtimedwait sigwait sigwaitinfo stime \
setdomainname sethostname symlinkat ulimit ustat vfork vhangup
vmsplice
+
+FILTER_OUT_DIRS += preadv pwritev
endif
==========
Then I can accomplish compiling LTP with android-ndk-r19c, and my configure as below.
==============
$ export NDK_TOOL_DIR=/path/to/android-ndk-r19c
$ ./configure --prefix=/tmp/ltp
CC=${NDK_TOOL_DIR}/toolchains/llvm/prebuilt/linux-x86_64/bin/clang
AR=${NDK_TOOL_DIR}/toolchains/llvm/prebuilt/linux-x86_64/bin/llvm-ar
CFLAGS="-isystem ${NDK_TOOL_DIR}/sysroot/usr/include -I
/media/Workspace/Softwares/Linux/toolschain/ndk/android-ndk-r19c/sysroot/usr/include
-I
/media/Workspace/Softwares/Linux/toolschain/ndk/android-ndk-r19c/sysroot/usr/include/aarch64-linux-android
-D__ANDROID__ -fmessage-length=0 -W -Wall -Wno-unused -Winit-self
-Wpointer-arith -no-canonical-prefixes -DNDEBUG -UDEBUG -fno-exceptions
-Wno-multichar -O2 -g -fno-strict-aliasing
-fdebug-prefix-map=/proc/self/cwd=
-D__compiler_offsetof=__builtin_offsetof -Werror=int-conversion
-Wno-reserved-id-macro -Wno-format-pedantic
-Wno-unused-command-line-argument -fcolor-diagnostics
-Wno-expansion-to-defined -Wno-zero-as-null-pointer-constant
-ffunction-sections -fdata-sections -fno-short-enums -funwind-tables
-fstack-protector-strong -Wa,--noexecstack -D_FORTIFY_SOURCE=2
-Wstrict-aliasing=2 -Werror=return-type -Werror=non-virtual-dtor
-Werror=address -Werror=sequence-point -Werror=date-time
-Werror=format-security -nostdlibinc -march=armv8-a -target
aarch64-linux-android
-B${NDK_TOOL_DIR}/toolchains/llvm/prebuilt/linux-x86_64/bin -std=gnu99
-Wall -fPIE -D_USING_LIBCXX -Werror=int-to-pointer-cast
-Werror=pointer-to-int-cast -Werror=address-of-temporary
-Werror=return-type -Wno-tautological-constant-compare
-Wno-null-pointer-arithmetic -Wno-enum-compare -Wno-enum-compare-switch
-MD -nostdlib -pthread" --host=aarch64-linux-gnu LDFLAGS="-nostdlib
-Wl,-dynamic-linker,/system/bin/linker64 -Wl,--gc-sections
-Wl,-z,nocopyreloc
-Wl,-rpath-link=${BUILD_PROJ_DIR}/out/target/product/e28/obj/lib
${NDK_TOOL_DIR}/toolchains/llvm/prebuilt/linux-x86_64/sysroot/usr/lib/aarch64-linux-android/28/crtbegin_dynamic.o
-Wl,--whole-archive -Wl,--no-whole-archive
${NDK_TOOL_DIR}/toolchains/llvm/prebuilt/linux-x86_64/lib64/clang/8.0.2/lib/linux/libclang_rt.ubsan_minimal-aarch64-android.a
${NDK_TOOL_DIR}/sysroot/usr/lib/aarch64-linux-android/libcompiler_rt-extras.a
${NDK_TOOL_DIR}/toolchains/llvm/prebuilt/linux-x86_64/aarch64-linux-android/lib64/libatomic.a
${NDK_TOOL_DIR}/toolchains/llvm/prebuilt/linux-x86_64/lib/gcc/aarch64-linux-android/4.9.x/libgcc.a
-Wl,-z,noexecstack -Wl,-z,relro -Wl,-z,now -Wl,--build-id=md5
-Wl,--warn-shared-textrel -Wl,--fatal-warnings
-Wl,--no-undefined-version -Wl,-m,aarch64_elf64_le_vec
-Wl,--hash-style=gnu -fuse-ld=gold -Wl,--icf=safe
-Wl,--exclude-libs,libclang_rt.ubsan_minimal-aarch64-android.a
-Wl,--no-undefined
-B${NDK_TOOL_DIR}/toolchains/llvm/prebuilt/linux-x86_64/bin
-L${NDK_TOOL_DIR}/toolchains/llvm/prebuilt/linux-x86_64/sysroot/usr/lib/aarch64-linux-android/28
-lc++ -lc -lm -ldl
${NDK_TOOL_DIR}/toolchains/llvm/prebuilt/linux-x86_64/sysroot/usr/lib/aarch64-linux-android/28/crtend_android.o
-target aarch64-linux-android"$ make ANDROID=1
Regards,
Zhengwang
Quoted 16 lines
Quoted 7 lines
1 error generated.
<builtin>: recipe for target 'preadv02' failed
make[4]: *** [preadv02] Error 1
make[4]: Leaving directory '/usr/local/google/home/sspatil/dev/ssd2/ltp/testcases/kernel/syscalls/preadv' ../../../include/mk/generic_trunk_target.inc:93: recipe for target 'all' failed
make[3]: *** [all] Error 2
make[3]: Leaving directory '/usr/local/google/home/sspatil/dev/ssd2/ltp/testcases/kernel/syscalls'Quoted 8 lines
../../include/mk/generic_trunk_target.inc:93: recipe for target 'all' failed make[2]: *** [all] Error 2
make[2]: Leaving directory '/usr/local/google/home/sspatil/dev/ssd2/ltp/testcases/kernel' ../include/mk/generic_trunk_target.inc:93: recipe for target 'all' failed make[1]: *** [all] Error 2
make[1]: Leaving directory '/usr/local/google/home/sspatil/dev/ssd2/ltp/testcases' Makefile:111: recipe for target 'testcases-all' failed
make: *** [testcases-all] Error 2
Quoted 2 lines
Kind regards,
Petr
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linux.it/pipermail/ltp/attachments/20190419/91ba37d3/attachment.html>
Hi Sandeep, Zhengwang,
Quoted 17 lines
Quoted 2 lines
I don't know why autotools fails preadv and pwritev detection with
AC_CHECK_FUNCS. I considered it as problem of my setup, but maybe it's an autotools bug.I also encountered this issue, but I temporarily walked around it by applying the following patch.
==================
$ git diff testcases/testcases/commands/ldd/datafiles/Makefile +0
-0@@ -30,7 +30,7 @@ CLEAN_TARGETS += *.obj $(LDD_SO_FILES)$(CC) $(CFLAGS) -shared -o $@ $^
Quoted 3 lines
lddfile.out: main.o $(LDD_SO_FILES)
- $(CC) $(CFLAGS) -O -o $@ $?
+ $(CC) $(CFLAGS) $(LDFLAGS) -O -o $@ $?
What is this change for?
Quoted 13 lines
INSTALL_DIR := testcases/data/ldd01
testcases/kernel/syscalls/Makefile +0
-0@@ -43,6 +43,8 @@ FILTER_OUT_DIRS += \
sched_getaffinitysched_setaffinitysendmsg setgroups setns sighold \
sigrelse sigsuspend sigtimedwait sigwait sigwaitinfo stime \
setdomainname sethostname symlinkat ulimit ustat vfork vhangup vmsplice
testcases/kernel/syscalls/Makefile
Quoted 2 lines
testcases/kernel/syscalls/Makefile +2
-0+ +FILTER_OUT_DIRS += preadv pwritev
Yes, that fixes build. But manually disabling features in config.h enables these tests to be build.
Kind regards,
Petr
Hi Petr,
-------- Original Message --------
From: Petr Vorel
Sent: Mon, 22 Apr 2019 00:10:30 +0200
To: Zhengwang Ruan
Cc: Sandeep Patil, Ltp, Steve Muckle
Subject: Re: [LTP] 回复:[PATCH v1] include/mk/env_post.mk : enable __ANDROID__ definition for Android build
Quoted 4 lines
Hi Sandeep, Zhengwang,
Quoted 2 lines
Quoted 2 lines
I don't know why autotools fails preadv and pwritev detection with
AC_CHECK_FUNCS. I considered it as problem of my setup, but maybe it's an autotools bug.
Quoted 15 lines
Quoted 10 lines
I also encountered this issue, but I temporarily walked around it by applying the following patch.
==================$ git diff testcases/testcases/commands/ldd/datafiles/Makefile +0
-0testcases/commands/ldd/datafiles/Makefile CLEAN_TARGETS += *.obj $(LDD_SO_FILES) · line 32
Quoted 4 lines
testcases/commands/ldd/datafiles/Makefile +1
-1@@ -30,7 +30,7 @@ CLEAN_TARGETS += *.obj $(LDD_SO_FILES) $(CC) $(CFLAGS) -shared -o $@ $^ lddfile.out: main.o $(LDD_SO_FILES) - $(CC) $(CFLAGS) -O -o $@ $? + $(CC) $(CFLAGS) $(LDFLAGS) -O -o $@ $?What is this change for?
This is a really long story. In my build, I need to use '-nostdlib /ndk/path/to/crtend_android.o /ndk/path/to/crtbegin_dynamic.o' as linker arguments to cross-compile, and I have to put these arguments into LDFLAGS since there were other errors occurred on linking *.o to executable if I put them into CFLAGS.
Right here, lddfile.out is an executable and obviously needed to be linked, therefore LDFLAGS must be required in my case.
Regards,
Zhengwang
Quoted 18 lines
Quoted 13 lines
testcases/commands/ldd/datafiles/Makefile +0
-0INSTALL_DIR := testcases/data/ldd01testcases/kernel/syscalls/Makefile +0
-0@@ -43,6 +43,8 @@ FILTER_OUT_DIRS += \ sched_getaffinity sched_setaffinity sendmsg setgroups setns sighold \ sigrelse sigsuspend sigtimedwait sigwait sigwaitinfo stime \ setdomainname sethostname symlinkat ulimit ustat vfork vhangupvmsplice
testcases/kernel/syscalls/Makefile
Quoted 2 lines
testcases/kernel/syscalls/Makefile +2
-0+ +FILTER_OUT_DIRS += preadv pwritevYes, that fixes build. But manually disabling features in
config.henables these tests to be build.
Quoted 2 lines
Kind regards,
Petr
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linux.it/pipermail/ltp/attachments/20190424/187172f2/attachment.html>
Hi Zhengwang,
Quoted 27 lines
-------- Original Message --------
From: Petr Vorel
Sent: Mon, 22 Apr 2019 00:10:30 +0200
To: Zhengwang Ruan
Cc: Sandeep Patil, Ltp, Steve Muckle
Subject: Re: [LTP] 回复:[PATCH v1]include/mk/env_post.mk: enable__ANDROID__definition for Android buildQuoted 4 lines
Hi Sandeep, Zhengwang,
Quoted 2 lines
Quoted 2 lines
I don't know why autotools fails preadv and pwritev detection with
AC_CHECK_FUNCS. I considered it as problem of my setup, but maybe it's an autotools bug.Quoted 15 lines
Quoted 10 lines
I also encountered this issue, but I temporarily walked around it by applying the following patch.
==================$ git diff testcases/testcases/commands/ldd/datafiles/Makefile +0
-0testcases/commands/ldd/datafiles/Makefile CLEAN_TARGETS += *.obj $(LDD_SO_FILES) · line 32
Quoted 4 lines
testcases/commands/ldd/datafiles/Makefile +1
-1@@ -30,7 +30,7 @@ CLEAN_TARGETS += *.obj $(LDD_SO_FILES) $(CC) $(CFLAGS) -shared -o $@ $^ lddfile.out: main.o $(LDD_SO_FILES) - $(CC) $(CFLAGS) -O -o $@ $? + $(CC) $(CFLAGS) $(LDFLAGS) -O -o $@ $?What is this change for?
Quoted 8 lines
This is a really long story. In my build, I need to use '-nostdlib /ndk/path/to/crtend_android.o /ndk/path/to/crtbegin_dynamic.o' as linker arguments to cross-compile, and I have to put these arguments into LDFLAGS since there were other errors occurred on linking *.o to executable if I put them into CFLAGS.
Right here, lddfile.out is an executable and obviously needed to be linked, therefore LDFLAGS must be required in my case.
Thanks for info. Feel free to send a patch with a long description.
Back to the patch: I'm willing to merge it, just wait if any from Sandeep or Steve (or any other android developer) will ack it.
Kind regards,
Petr
-------- Original Message --------
From: Sandeep Patil
Sent: Thu, 11 Apr 2019 09:27:24 -0700
To: Petr Vorel
Cc: Ltp
Subject: Re: [LTP] 回复:[PATCH v1] include/mk/env_post.mk: enable __ANDROID__ definition for Android b uild
Quoted 34 lines · Petr Vorel
Quoted 25 lines
Hi Sandeep, Steve,
Quoted 23 lines
Could anyone help me to review this patch? This is useful to cross-compile ltp for Android without specifying '-D__ANDROID__' in configure step. Thanks!
Zhengwang------------------------------------------------------------------
发件人:Zhengwang Ruan <[email protected]>
发送时间:2019年4月9日(星期二) 12:05
收件人:ltp <[email protected]>
抄 送:Zhengwang Ruan <[email protected]>
主 题:[PATCH v1]include/mk/env_post.mk: enable__ANDROID__definition for Android build Enable the__ANDROID__definition by default for Android build, otherwise we should manually enable it by configure command.Signed-off-by Zhengwang Ruan <[email protected]>--- include/mk/env_post.mk | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)include/mk/env_post.mk +0
-0@@ -44,7 +44,7 @@ endif ifeq ($(ANDROID),1) # There are many undeclared functions, it's best not to accidentally overlookQuoted 8 lines
include/mk/env_post.mk endif · line 47
Quoted 3 lines
include/mk/env_post.mk +1
-1# them. -CFLAGS += -Werror-implicit-function-declaration +CFLAGS += -Werror-implicit-function-declaration -D__ANDROID__Would adding
__ANDROID__definition help building LTP with distro cross-compiler? (i.e. no NDK, no inside AOSP tree)?
I guess so and therefore, I'd be for this change.
I see some projects added it as well (busybox [1] [2], dnsmasq [3]).
BTW: where is defined for AOSP and NDK?
Quoted 3 lines
I think I've used
__ANDROID__and__BIONIC__in the past for android-specific things. Honestly, I haven't tried to build for Android outside of AOSP tree yet. I am happy to test this if you tell me how? (sorry for that.)
Hi Sandeep,
You can try the steps below to make a build outside AOSP. In my case, it works fine.
$ autoreconf --install
$ ./configure --prefix=/tmp/ltp
CC=${BUILD_PROJ_DIR}/prebuilts/clang/host/linux-x86/clang-4691093/bin/clang
AR=${BUILD_PROJ_DIR}/prebuilts/clang/host/linux-x86/clang-4691093/bin/llvm-ar
CFLAGS="-I ${BUILD_PROJ_DIR}/libnativehelper/include_jni -I
${BUILD_PROJ_DIR}/system/core/include -I
${BUILD_PROJ_DIR}/system/media/audio/include -I
${BUILD_PROJ_DIR}/hardware/libhardware/include -I
${BUILD_PROJ_DIR}/hardware/libhardware_legacy/include -I
${BUILD_PROJ_DIR}/hardware/ril/include -I
${BUILD_PROJ_DIR}/libnativehelper/include -I
${BUILD_PROJ_DIR}/frameworks/native/include -I
${BUILD_PROJ_DIR}/frameworks/native/opengl/include -I
${BUILD_PROJ_DIR}/frameworks/av/include -isystem
${BUILD_PROJ_DIR}/out/target/product/${PRODUCT_NAME}/obj/include
-isystem ${BUILD_PROJ_DIR}/bionic/libc/include -isystem
${BUILD_PROJ_DIR}/bionic/libc/kernel/uapi -isystem
${BUILD_PROJ_DIR}/bionic/libc/kernel/uapi/asm-arm64 -isystem
${BUILD_PROJ_DIR}/bionic/libc/kernel/android/scsi -isystem
${BUILD_PROJ_DIR}/bionic/libc/kernel/android/uapi -target
aarch64-linux-android
-B${BUILD_PROJ_DIR}/prebuilts/gcc/linux-x86/aarch64/aarch64-linux-android-4.9/aarch64-linux-android/bin
-pthread -nostdlib -D__ANDROID__" --host=aarch64-linux-gnu
LDFLAGS="-nostdlib -Wl,-dynamic-linker,/system/bin/linker64
-Wl,--gc-sections -Wl,-z,nocopyreloc
-Wl,-rpath-link=${BUILD_PROJ_DIR}/out/target/product/${PRODUCT_NAME}/obj/lib
${BUILD_PROJ_DIR}/out/target/product/${PRODUCT_NAME}/obj/lib/crtbegin_dynamic.o
-Wl,--whole-archive -Wl,--no-whole-archive
${BUILD_PROJ_DIR}/out/target/product/${PRODUCT_NAME}/obj/STATIC_LIBRARIES/libclang_rt.ubsan_minimal-aarch64-android_intermediates/libclang_rt.ubsan_minimal-aarch64-android.a
${BUILD_PROJ_DIR}/out/target/product/${PRODUCT_NAME}/obj/STATIC_LIBRARIES/libcompiler_rt-extras_intermediates/libcompiler_rt-extras.a
${BUILD_PROJ_DIR}/out/target/product/${PRODUCT_NAME}/obj/STATIC_LIBRARIES/libatomic_intermediates/libatomic.a
${BUILD_PROJ_DIR}/out/target/product/${PRODUCT_NAME}/obj/STATIC_LIBRARIES/libgcc_intermediates/libgcc.a
-Wl,-z,noexecstack -Wl,-z,relro -Wl,-z,now -Wl,--build-id=md5
-Wl,--warn-shared-textrel -Wl,--fatal-warnings
-Wl,--no-undefined-version -Wl,-m,aarch64_elf64_le_vec
-Wl,--hash-style=gnu -fuse-ld=gold -Wl,--icf=safe
-Wl,--exclude-libs,libclang_rt.ubsan_minimal-aarch64-android.a
-Wl,--no-undefined
-B${BUILD_PROJ_DIR}/prebuilts/gcc/linux-x86/aarch64/aarch64-linux-android-4.9/aarch64-linux-android/bin
-L${BUILD_PROJ_DIR}/out/target/product/${PRODUCT_NAME}/obj/lib -lc++ -lc
-lm -ldl
${BUILD_PROJ_DIR}/out/target/product/${PRODUCT_NAME}/obj/lib/crtend_android.o"$ make ANDROID=1
BTW, ${BUILD_PROJ_DIR} refers to the path of your AOSP project, and ${PRODUCT_NAME} refers to the product nickname inside your organization. you need to export these two variables before launch the above configure.
But unfortunately, you will encounter other build errors, especially those caused by -Werror-implicit-function-declaration. But I think it should be enough to tell the deference between __ANDROID__ and ANDROID.
Good luck! :-)
Zhengwang
Quoted 5 lines
Steve will also be back by 4/15 and he would know this more than I do. So, Ideally, if we can wait for him it will be great. In the mean time, I can test..
Quoted 9 lines
FWIW, I want to be able to do this as well. I know its a lot of pain to have to checkout 100s of gigabytes of AOSP tree in order to build LTP for Android right now. So, let me know how I can test and compare.
<snip>
- ssp
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linux.it/pipermail/ltp/attachments/20190412/ccb40a26/attachment.html>
PATCH 1/1 include/mk/env_post.mk: enable __ANDROID__ definition for Android build
In reply to Zhengwang RuanHi Steve, Sandeep,
Quoted 13 lines
Enable the
__ANDROID__definition by default for Android build, otherwise we should manually enable it by configure command.Signed-off-by Zhengwang Ruan <[email protected]>--- include/mk/env_post.mk | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)include/mk/env_post.mk +0
-0
include/mk/env_post.mk endif · line 50
Quoted 8 lines
include/mk/env_post.mk +1
-1@@ -44,7 +44,7 @@ endif ifeq ($(ANDROID),1) # There are many undeclared functions, it's best not to accidentally overlook # them. -CFLAGS += -Werror-implicit-function-declaration +CFLAGS += -Werror-implicit-function-declaration -D__ANDROID__ LDFLAGS += -L$(top_builddir)/lib/android_libpthread LDFLAGS += -L$(top_builddir)/lib/android_librt
Please, when you have time, can you please have a look into this?
IMHO the question is whether code guarded by __ANDROID__ (lib/tst_kernel.c, rt_tgsigqueueinfo01.c, etc.) works as expected in NDK build (as in aosp you use definitions in Android.bp anyway).
According to [1] -D__ANDROID__ is defined by toolchain (I understand it by both aosp build and builds using NDK), so it shouldn't be needed.
Not sure if it makes sense to build for android with non android toolchain (i.e. no aosp nor NDK, use generic arm cross compile toolchain or standard gcc/clang in case of intel platform). If yes, than this patch would be useful, as generic toolchain obviously don't define __ANDROID__.
Kind regards,
Petr
[1] https://groups.google.com/forum/#!topic/android-ndk/cf9_f1SLXls
PATCH 1/1 include/mk/env_post.mk: enable __ANDROID__ definition for Android build
Hi Petr, Zhengwang,
Quoted 2 lines · Petr Vorel
Hi Steve, Sandeep,
...
Quoted 9 lines
Please, when you have time, can you please have a look into this?
IMHO the question is whether code guarded by
__ANDROID__(lib/tst_kernel.c,rt_tgsigqueueinfo01.c, etc.) works as expected in NDK build (as in aosp you use definitions in Android.bp anyway).According to [1] -D__ANDROID__ is defined by toolchain (I understand it by both aosp build and builds using NDK), so it shouldn't be needed.
Using Sandeep's instructions and with your latest merged fixes I was able to build unmodified upstream with the NDK. I just tried running one of the test binaries on my device to sanity check it.
I put in an #error if __ANDROID__ is defined and it does look to be set by the NDK compiler. FWIW I am using Android API level 26, the minimum which provides hasmntopt. This gets configured in the "Set up ndk toolchains for autoconf" step Sandeep mentioned by setting the CC and CXX env vars appropriately.
Zhengwang can you try the steps Sandeep mentioned (with the API level tweak I mentioned above)?
Quoted 4 lines
Not sure if it makes sense to build for android with non android toolchain (i.e. no aosp nor NDK, use generic arm cross compile toolchain or standard gcc/clang in case of intel platform). If yes, than this patch would be useful, as generic toolchain obviously don't define
__ANDROID__.
Zhengwang had said
Quoted 3 lines
In the case of cross-compilation, I think it still makes sense, since
we can use the same toolchain, such as clang, and build parameters (the specify the include headers and libs, etc) as AOSP.
but it's not clear to me since you need to have bionic around anyway, and at that point, why not just use the Android toolchain?
thanks,
Steve
PATCH 1/1 include/mk/env_post.mk: enable __ANDROID__ definition for Android build
Hi Steve, Zhengwang,
Quoted 9 lines
Using Sandeep's instructions and with your latest merged fixes I was able to build unmodified upstream with the NDK. I just tried running one of the test binaries on my device to sanity check it.
I put in an #error if
__ANDROID__is defined and it does look to be set by the NDK compiler. FWIW I am using Android API level 26, the minimum which provides hasmntopt. This gets configured in the "Set up ndk toolchains for autoconf" step Sandeep mentioned by setting the CC and CXX env vars appropriately.
Thanks for testing. IMHO this is a simple proof that this patch (-D__ANDROID__) is not needed.
...
Quoted 6 lines
Quoted 3 lines
In the case of cross-compilation, I think it still makes sense, since
we can use the same toolchain, such as clang, and build parameters (the specify the include headers and libs, etc) as AOSP.but it's not clear to me since you need to have bionic around anyway, and at that point, why not just use the Android toolchain?
No, you're right, that does not make sense.
Quoted 2 lines
thanks,
Steve
Kind regards,
Petr
PATCH 1/1 include/mk/env_post.mk: enable __ANDROID__ definition for Android build
In reply to Steve MuckleHi Steve,
-------- Original Message --------
From: Steve Muckle
Sent: Wed, 24 Apr 2019 12:23:01 -0700
To: Petr Vorel, Zhengwang Ruan
Cc: Sandeep Patil, Ltp, Kernel-team
Subject: Re: [LTP] [PATCH v1] include/mk/env_post.mk: enable __ANDROID__ definition for Android build
Quoted 23 lines
Hi Petr, Zhengwang,
Quoted 2 lines · Petr Vorel
Hi Steve, Sandeep,
...
Quoted 9 lines
Please, when you have time, can you please have a look into this?
IMHO the question is whether code guarded by
__ANDROID__(lib/tst_kernel.c,rt_tgsigqueueinfo01.c, etc.) works as expected in NDK build (as in aosp you use
definitions in Android.bp anyway).Quoted 3 lines
According to [1] -D__ANDROID__ is defined by toolchain (I understand it by both
aosp build and builds using NDK), so it shouldn't be needed.Using Sandeep's instructions and with your latest merged fixes I was able to build unmodified upstream with the NDK. I just tried running one of the test binaries on my device to sanity check it.
Quoted 8 lines
I put in an #error if
__ANDROID__is defined and it does look to be set by the NDK compiler. FWIW I am using Android API level 26, the minimum which provides hasmntopt. This gets configured in the "Set up ndk toolchains for autoconf" step Sandeep mentioned by setting the CC and CXX env vars appropriately.Zhengwang can you try the steps Sandeep mentioned (with the API level tweak I mentioned above)?
Yes, I tried the steps as Sandeep mentioned and also put an "#error" if __ANDROID__ was defined, it proves NDK has defined it.
Thanks you guys very much for letting me know this. :-)
Quoted 9 lines
Quoted 8 lines
Not sure if it makes sense to build for android with non android toolchain (i.e.
no aosp nor NDK, use generic arm cross compile toolchain or standard gcc/clang
in case of intel platform). If yes, than this patch would be useful, as generic
toolchain obviously don't define__ANDROID__.
Quoted 8 lines
Zhengwang had said
Quoted 3 lines
In the case of cross-compilation, I think it still makes sense, since we can use the same toolchain, such as clang, and build parameters (the specify the include headers and libs, etc) as AOSP.
but it's not clear to me since you need to have bionic around anyway, and at that point, why not just use the Android toolchain?
Yes, NDK should always be the best in the situation of cross-compiling for android. Thanks again!
Regards,
Zhengwang
Quoted 4 lines
thanks,
Steve
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linux.it/pipermail/ltp/attachments/20190425/7ef04acf/attachment-0001.html>