Suggested patches for Linux 4.9 issues

[l***@gmail.com]

Hi,

I personally don't use Android x86, but I maintain an Android port for a
x86 device using a Linux 4.9 kernel just like Android x86. Recently I had
to backport a few commits from AOSP's master branch to fix issues with the
newer kernel together with some Android components and I thought I'd
suggest them here if you haven't applied them already.

Note: I didn't test them on Android x86 but I assume it's affected by the
same issues as my device.

*Patch for ART crash in Linux 4.9.34+:*
There was a change in Linux 4.9.34 (and probably newer kernels too) that
causes a crash in ART when booting the device. It's related to some memory
management change that was back-ported to the 4.9 kernel. The commit below
fixes it for me.

ART: Change main-thread thread paging schemehttps://android.googlesource.com/platform/art/+/6c63f863f9dbf734732ff837c1e772b3482e59ff


*Patch for system_netd issues:*
I've had a short glance at the netd hack patch that was posted here a few
times and it looks related, but for me back-porting two commits fixed it
properly. It's related to a change that was made when upstreaming a change
to some network routing code, which requires a change in system_netd:

Set both legacy and new UID routing attributes.
https://android.googlesource.com/platform/system/netd/+/2b078678aafceeefea6a70e96ab8ddefe515d027z


Use new-style UID routing.https://android.googlesource.com/platform/system/netd/+/22c24ebd283d6e8fb782b34975417308839d4ad7


On the Android 4.9 kernel from Google, the netd issue wasn't as obvious
because it only happens when cleaning up a network interface (e.g. when
switching to VPN) with a really weird error. That's due to a kernel bug
that has been only fixed in newer kernel versions. After I back-ported the
patch below there was an error when initializing the network interface
(that was hidden before). This is fixed by the two patches I posted above.

fib_rules: fix error return codehttps://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git/commit/?id=adeb45cbb5057731ce9c47aad93756135d7947bf


I hope they are helpful for you. :)
Lambdadroid

[youling 257]

Buti from kernel source to solved art
crashhttps://github.com/youling257/android-4.9/commits/update?author=youling257
https://github.com/youling257/android-4.9/commits/master?after=149295eb63a3a58e458156cb1125f09d1fe86981+104

revert "mm... "i won't to build 50GB+ Android sourcei build kernel
sourceuse custom kernel to boot Phoenix/lineage/Bliss/Androidx86i only
can wating the iso.

圚 2017幎8月23日星期䞉 UTC+8䞋午7:47:37lambdadroid写道

Post by l***@gmail.com
Hi,
I personally don't use Android x86, but I maintain an Android port for a
x86 device using a Linux 4.9 kernel just like Android x86. Recently I had
to backport a few commits from AOSP's master branch to fix issues with the
newer kernel together with some Android components and I thought I'd
suggest them here if you haven't applied them already.
Note: I didn't test them on Android x86 but I assume it's affected by the
same issues as my device.
*Patch for ART crash in Linux 4.9.34+:*
There was a change in Linux 4.9.34 (and probably newer kernels too) that
causes a crash in ART when booting the device. It's related to some memory
management change that was back-ported to the 4.9 kernel. The commit below
fixes it for me.
ART: Change main-thread thread paging schemehttps://android.googlesource.com/platform/art/+/6c63f863f9dbf734732ff837c1e772b3482e59ff
*Patch for system_netd issues:*
I've had a short glance at the netd hack patch that was posted here a few
times and it looks related, but for me back-porting two commits fixed it
properly. It's related to a change that was made when upstreaming a change
Set both legacy and new UID routing attributes.
https://android.googlesource.com/platform/system/netd/+/2b078678aafceeefea6a70e96ab8ddefe515d027z
Use new-style UID routing.https://android.googlesource.com/platform/system/netd/+/22c24ebd283d6e8fb782b34975417308839d4ad7
On the Android 4.9 kernel from Google, the netd issue wasn't as obvious
because it only happens when cleaning up a network interface (e.g. when
switching to VPN) with a really weird error. That's due to a kernel bug
that has been only fixed in newer kernel versions. After I back-ported the
patch below there was an error when initializing the network interface
(that was hidden before). This is fixed by the two patches I posted above.
fib_rules: fix error return codehttps://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git/commit/?id=adeb45cbb5057731ce9c47aad93756135d7947bf
I hope they are helpful for you. :)
Lambdadroid

[lambdadroid]

Reverting the kernel patches that have caused the issue only avoids the
problem, it doesn't solve it. The patches improve kernel hardening, so
reverting them is not the best solution in my opinion.

It's an issue that should be fixed in ART, not in the kernel.

Buti from kernel source to solved art crash
https://github.com/youling257/android-4.9/commits/update?author=youling257
https://github.com/youling257/android-4.9/commits/master?after=149295eb63a3a58e458156cb1125f09d1fe86981+104
revert "mm... "i won't to build 50GB+ Android sourcei build kernel
sourceuse custom kernel to boot Phoenix/lineage/Bliss/Androidx86i only
can wating the iso.
圚 2017幎8月23日星期䞉 UTC+8䞋午7:47:37lambdadroid写道

Post by l***@gmail.com
Hi,
I personally don't use Android x86, but I maintain an Android port for a
x86 device using a Linux 4.9 kernel just like Android x86. Recently I had
to backport a few commits from AOSP's master branch to fix issues with the
newer kernel together with some Android components and I thought I'd
suggest them here if you haven't applied them already.
Note: I didn't test them on Android x86 but I assume it's affected by the
same issues as my device.
*Patch for ART crash in Linux 4.9.34+:*
There was a change in Linux 4.9.34 (and probably newer kernels too) that
causes a crash in ART when booting the device. It's related to some memory
management change that was back-ported to the 4.9 kernel. The commit below
fixes it for me.
ART: Change main-thread thread paging schemehttps://android.googlesource.com/platform/art/+/6c63f863f9dbf734732ff837c1e772b3482e59ff
*Patch for system_netd issues:*
I've had a short glance at the netd hack patch that was posted here a few
times and it looks related, but for me back-porting two commits fixed it
properly. It's related to a change that was made when upstreaming a change
Set both legacy and new UID routing attributes.
https://android.googlesource.com/platform/system/netd/+/2b078678aafceeefea6a70e96ab8ddefe515d027z
Use new-style UID routing.https://android.googlesource.com/platform/system/netd/+/22c24ebd283d6e8fb782b34975417308839d4ad7
On the Android 4.9 kernel from Google, the netd issue wasn't as obvious
because it only happens when cleaning up a network interface (e.g. when
switching to VPN) with a really weird error. That's due to a kernel bug
that has been only fixed in newer kernel versions. After I back-ported the
patch below there was an error when initializing the network interface
(that was hidden before). This is fixed by the two patches I posted above.
fib_rules: fix error return codehttps://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git/commit/?id=adeb45cbb5057731ce9c47aad93756135d7947bf
I hope they are helpful for you. :)
Lambdadroid

[youling 257]

if 4.9 kernel add this
patchhttps://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git/commit/?id=adeb45cbb5057731ce9c47aad93756135d7947bf
nougat vpn should be working ?
i haven't test vpn on 4.12/4.13 kernelhere has a netd hack for 4.12/4.13
kernel

圚 2017幎8月23日星期䞉 UTC+8䞋午8:18:39lambdadroid写道

Post by lambdadroid
Reverting the kernel patches that have caused the issue only avoids the
problem, it doesn't solve it. The patches improve kernel hardening, so
reverting them is not the best solution in my opinion.
It's an issue that should be fixed in ART, not in the kernel.

Buti from kernel source to solved art crash
https://github.com/youling257/android-4.9/commits/update?author=youling257
https://github.com/youling257/android-4.9/commits/master?after=149295eb63a3a58e458156cb1125f09d1fe86981+104
revert "mm... "i won't to build 50GB+ Android sourcei build kernel
sourceuse custom kernel to boot Phoenix/lineage/Bliss/Androidx86i only
can wating the iso.
圚 2017幎8月23日星期䞉 UTC+8䞋午7:47:37lambdadroid写道

Post by l***@gmail.com
Hi,
I personally don't use Android x86, but I maintain an Android port for a
x86 device using a Linux 4.9 kernel just like Android x86. Recently I had
to backport a few commits from AOSP's master branch to fix issues with the
newer kernel together with some Android components and I thought I'd
suggest them here if you haven't applied them already.
Note: I didn't test them on Android x86 but I assume it's affected by
the same issues as my device.
*Patch for ART crash in Linux 4.9.34+:*
There was a change in Linux 4.9.34 (and probably newer kernels too) that
causes a crash in ART when booting the device. It's related to some memory
management change that was back-ported to the 4.9 kernel. The commit below
fixes it for me.
ART: Change main-thread thread paging schemehttps://android.googlesource.com/platform/art/+/6c63f863f9dbf734732ff837c1e772b3482e59ff
*Patch for system_netd issues:*
I've had a short glance at the netd hack patch that was posted here a
few times and it looks related, but for me back-porting two commits fixed
it properly. It's related to a change that was made when upstreaming a
Set both legacy and new UID routing attributes.
https://android.googlesource.com/platform/system/netd/+/2b078678aafceeefea6a70e96ab8ddefe515d027z
Use new-style UID routing.https://android.googlesource.com/platform/system/netd/+/22c24ebd283d6e8fb782b34975417308839d4ad7
On the Android 4.9 kernel from Google, the netd issue wasn't as obvious
because it only happens when cleaning up a network interface (e.g. when
switching to VPN) with a really weird error. That's due to a kernel bug
that has been only fixed in newer kernel versions. After I back-ported the
patch below there was an error when initializing the network interface
(that was hidden before). This is fixed by the two patches I posted above.
fib_rules: fix error return codehttps://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git/commit/?id=adeb45cbb5057731ce9c47aad93756135d7947bf
I hope they are helpful for you. :)
Lambdadroid

[lambdadroid]

The kernel patch doesn't do anything except for making the problem visible,
it won't make VPN working.

For me, VPN was working after I applied the two netd patches I posted
above. It's possible that it's also working using the netd hack but I
haven't tested that.

if 4.9 kernel add this patch
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git/commit/?id=adeb45cbb5057731ce9c47aad93756135d7947bf
nougat vpn should be working ?
i haven't test vpn on 4.12/4.13 kernelhere has a netd hack for 4.12/4.13
kernel
圚 2017幎8月23日星期䞉 UTC+8䞋午8:18:39lambdadroid写道

Post by lambdadroid
Reverting the kernel patches that have caused the issue only avoids the
problem, it doesn't solve it. The patches improve kernel hardening, so
reverting them is not the best solution in my opinion.
It's an issue that should be fixed in ART, not in the kernel.

Buti from kernel source to solved art crash
https://github.com/youling257/android-4.9/commits/update?author=youling257
https://github.com/youling257/android-4.9/commits/master?after=149295eb63a3a58e458156cb1125f09d1fe86981+104
revert "mm... "i won't to build 50GB+ Android sourcei build kernel
sourceuse custom kernel to boot Phoenix/lineage/Bliss/Androidx86i only
can wating the iso.
圚 2017幎8月23日星期䞉 UTC+8䞋午7:47:37lambdadroid写道

Post by l***@gmail.com
Hi,
I personally don't use Android x86, but I maintain an Android port for
a x86 device using a Linux 4.9 kernel just like Android x86. Recently I had
to backport a few commits from AOSP's master branch to fix issues with the
newer kernel together with some Android components and I thought I'd
suggest them here if you haven't applied them already.
Note: I didn't test them on Android x86 but I assume it's affected by
the same issues as my device.
*Patch for ART crash in Linux 4.9.34+:*
There was a change in Linux 4.9.34 (and probably newer kernels too)
that causes a crash in ART when booting the device. It's related to some
memory management change that was back-ported to the 4.9 kernel. The commit
below fixes it for me.
ART: Change main-thread thread paging schemehttps://android.googlesource.com/platform/art/+/6c63f863f9dbf734732ff837c1e772b3482e59ff
*Patch for system_netd issues:*
I've had a short glance at the netd hack patch that was posted here a
few times and it looks related, but for me back-porting two commits fixed
it properly. It's related to a change that was made when upstreaming a
Set both legacy and new UID routing attributes.
https://android.googlesource.com/platform/system/netd/+/2b078678aafceeefea6a70e96ab8ddefe515d027z
Use new-style UID routing.https://android.googlesource.com/platform/system/netd/+/22c24ebd283d6e8fb782b34975417308839d4ad7
On the Android 4.9 kernel from Google, the netd issue wasn't as obvious
because it only happens when cleaning up a network interface (e.g. when
switching to VPN) with a really weird error. That's due to a kernel bug
that has been only fixed in newer kernel versions. After I back-ported the
patch below there was an error when initializing the network interface
(that was hidden before). This is fixed by the two patches I posted above.
fib_rules: fix error return codehttps://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git/commit/?id=adeb45cbb5057731ce9c47aad93756135d7947bf
I hope they are helpful for you. :)
Lambdadroid

[youling 257]

yes4.13 kernel vpn not working.
the Phoenix os 2.2.0 fixed vpn problemmay be them used
https://android.googlesource.com/platform/system/netd/+/22c24ebd283d6e8fb782b34975417308839d4ad7
https://android.googlesource.com/platform/system/netd/+/2b078678aafceeefea6a70e96ab8ddefe515d027

圚 2017幎8月23日星期䞉 UTC+8䞋午8:38:36lambdadroid写道

Post by lambdadroid
The kernel patch doesn't do anything except for making the problem
visible, it won't make VPN working.
For me, VPN was working after I applied the two netd patches I posted
above. It's possible that it's also working using the netd hack but I
haven't tested that.

if 4.9 kernel add this patch
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git/commit/?id=adeb45cbb5057731ce9c47aad93756135d7947bf
nougat vpn should be working ?
i haven't test vpn on 4.12/4.13 kernelhere has a netd hack for 4.12/4.13
kernel
圚 2017幎8月23日星期䞉 UTC+8䞋午8:18:39lambdadroid写道

Post by lambdadroid
Reverting the kernel patches that have caused the issue only avoids the
problem, it doesn't solve it. The patches improve kernel hardening, so
reverting them is not the best solution in my opinion.
It's an issue that should be fixed in ART, not in the kernel.

Buti from kernel source to solved art crash
https://github.com/youling257/android-4.9/commits/update?author=youling257
https://github.com/youling257/android-4.9/commits/master?after=149295eb63a3a58e458156cb1125f09d1fe86981+104
revert "mm... "i won't to build 50GB+ Android sourcei build kernel
sourceuse custom kernel to boot Phoenix/lineage/Bliss/Androidx86i only
can wating the iso.
圚 2017幎8月23日星期䞉 UTC+8䞋午7:47:37lambdadroid写道

Post by l***@gmail.com
Hi,
I personally don't use Android x86, but I maintain an Android port for
a x86 device using a Linux 4.9 kernel just like Android x86. Recently I had
to backport a few commits from AOSP's master branch to fix issues with the
newer kernel together with some Android components and I thought I'd
suggest them here if you haven't applied them already.
Note: I didn't test them on Android x86 but I assume it's affected by
the same issues as my device.
*Patch for ART crash in Linux 4.9.34+:*
There was a change in Linux 4.9.34 (and probably newer kernels too)
that causes a crash in ART when booting the device. It's related to some
memory management change that was back-ported to the 4.9 kernel. The commit
below fixes it for me.
ART: Change main-thread thread paging schemehttps://android.googlesource.com/platform/art/+/6c63f863f9dbf734732ff837c1e772b3482e59ff
*Patch for system_netd issues:*
I've had a short glance at the netd hack patch that was posted here a
few times and it looks related, but for me back-porting two commits fixed
it properly. It's related to a change that was made when upstreaming a
Set both legacy and new UID routing attributes.
https://android.googlesource.com/platform/system/netd/+/2b078678aafceeefea6a70e96ab8ddefe515d027z
Use new-style UID routing.https://android.googlesource.com/platform/system/netd/+/22c24ebd283d6e8fb782b34975417308839d4ad7
On the Android 4.9 kernel from Google, the netd issue wasn't as
obvious because it only happens when cleaning up a network interface (e.g.
when switching to VPN) with a really weird error. That's due to a kernel
bug that has been only fixed in newer kernel versions. After I back-ported
the patch below there was an error when initializing the network interface
(that was hidden before). This is fixed by the two patches I posted above.
fib_rules: fix error return codehttps://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git/commit/?id=adeb45cbb5057731ce9c47aad93756135d7947bf
I hope they are helpful for you. :)
Lambdadroid

[lambdadroid]

I'm not sure. As I mentioned, I haven't tested this on Android x86, I use
this on a ROM that is mostly just LineageOS 14.1 with a few fixes to make
things work on a specific x86 device.

VPN worked for me with Linux 4.9 kernel, the 2 netd patches and OpenVPN for
Android. I've also seen other VPN apps that are not working properly,
though.

Post by youling 257
yes4.13 kernel vpn not working.
the Phoenix os 2.2.0 fixed vpn problemmay be them used
https://android.googlesource.com/platform/system/netd/+/22c24ebd283d6e8fb782b34975417308839d4ad7
https://android.googlesource.com/platform/system/netd/+/2b078678aafceeefea6a70e96ab8ddefe515d027
圚 2017幎8月23日星期䞉 UTC+8䞋午8:38:36lambdadroid写道

Post by lambdadroid
The kernel patch doesn't do anything except for making the problem
visible, it won't make VPN working.
For me, VPN was working after I applied the two netd patches I posted
above. It's possible that it's also working using the netd hack but I
haven't tested that.

if 4.9 kernel add this patch
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git/commit/?id=adeb45cbb5057731ce9c47aad93756135d7947bf
nougat vpn should be working ?
i haven't test vpn on 4.12/4.13 kernelhere has a netd hack for
4.12/4.13 kernel
圚 2017幎8月23日星期䞉 UTC+8䞋午8:18:39lambdadroid写道

Post by lambdadroid
Reverting the kernel patches that have caused the issue only avoids the
problem, it doesn't solve it. The patches improve kernel hardening, so
reverting them is not the best solution in my opinion.
It's an issue that should be fixed in ART, not in the kernel.

Buti from kernel source to solved art crash
https://github.com/youling257/android-4.9/commits/update?author=youling257
https://github.com/youling257/android-4.9/commits/master?after=149295eb63a3a58e458156cb1125f09d1fe86981+104
revert "mm... "i won't to build 50GB+ Android sourcei build kernel
sourceuse custom kernel to boot Phoenix/lineage/Bliss/Androidx86i only
can wating the iso.
圚 2017幎8月23日星期䞉 UTC+8䞋午7:47:37lambdadroid写道

Post by l***@gmail.com
Hi,
I personally don't use Android x86, but I maintain an Android port
for a x86 device using a Linux 4.9 kernel just like Android x86. Recently I
had to backport a few commits from AOSP's master branch to fix issues with
the newer kernel together with some Android components and I thought I'd
suggest them here if you haven't applied them already.
Note: I didn't test them on Android x86 but I assume it's affected by
the same issues as my device.
*Patch for ART crash in Linux 4.9.34+:*
There was a change in Linux 4.9.34 (and probably newer kernels too)
that causes a crash in ART when booting the device. It's related to some
memory management change that was back-ported to the 4.9 kernel. The commit
below fixes it for me.
ART: Change main-thread thread paging schemehttps://android.googlesource.com/platform/art/+/6c63f863f9dbf734732ff837c1e772b3482e59ff
*Patch for system_netd issues:*
I've had a short glance at the netd hack patch that was posted here a
few times and it looks related, but for me back-porting two commits fixed
it properly. It's related to a change that was made when upstreaming a
Set both legacy and new UID routing attributes.
https://android.googlesource.com/platform/system/netd/+/2b078678aafceeefea6a70e96ab8ddefe515d027z
Use new-style UID routing.https://android.googlesource.com/platform/system/netd/+/22c24ebd283d6e8fb782b34975417308839d4ad7
On the Android 4.9 kernel from Google, the netd issue wasn't as
obvious because it only happens when cleaning up a network interface (e.g.
when switching to VPN) with a really weird error. That's due to a kernel
bug that has been only fixed in newer kernel versions. After I back-ported
the patch below there was an error when initializing the network interface
(that was hidden before). This is fixed by the two patches I posted above.
fib_rules: fix error return codehttps://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git/commit/?id=adeb45cbb5057731ce9c47aad93756135d7947bf
I hope they are helpful for you. :)
Lambdadroid

[youling 257]

I just need test ShadowsocksR vpn modevpn not workingshadowsocksR NAT
mode workingbut v2ray only has vpn mode.
https://github.com/V2Ray-Android/Actinium/releases

圚 2017幎8月23日星期䞉 UTC+8䞋午8:52:52lambdadroid写道

Post by lambdadroid
I'm not sure. As I mentioned, I haven't tested this on Android x86, I use
this on a ROM that is mostly just LineageOS 14.1 with a few fixes to make
things work on a specific x86 device.
VPN worked for me with Linux 4.9 kernel, the 2 netd patches and OpenVPN
for Android. I've also seen other VPN apps that are not working properly,
though.

Post by youling 257
yes4.13 kernel vpn not working.
the Phoenix os 2.2.0 fixed vpn problemmay be them used
https://android.googlesource.com/platform/system/netd/+/22c24ebd283d6e8fb782b34975417308839d4ad7
https://android.googlesource.com/platform/system/netd/+/2b078678aafceeefea6a70e96ab8ddefe515d027
圚 2017幎8月23日星期䞉 UTC+8䞋午8:38:36lambdadroid写道

Post by lambdadroid
The kernel patch doesn't do anything except for making the problem
visible, it won't make VPN working.
For me, VPN was working after I applied the two netd patches I posted
above. It's possible that it's also working using the netd hack but I
haven't tested that.

if 4.9 kernel add this patch
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git/commit/?id=adeb45cbb5057731ce9c47aad93756135d7947bf
nougat vpn should be working ?
i haven't test vpn on 4.12/4.13 kernelhere has a netd hack for
4.12/4.13 kernel
圚 2017幎8月23日星期䞉 UTC+8䞋午8:18:39lambdadroid写道

Post by lambdadroid
Reverting the kernel patches that have caused the issue only avoids
the problem, it doesn't solve it. The patches improve kernel hardening, so
reverting them is not the best solution in my opinion.
It's an issue that should be fixed in ART, not in the kernel.

Buti from kernel source to solved art crash
https://github.com/youling257/android-4.9/commits/update?author=youling257
https://github.com/youling257/android-4.9/commits/master?after=149295eb63a3a58e458156cb1125f09d1fe86981+104
revert "mm... "i won't to build 50GB+ Android sourcei build kernel
sourceuse custom kernel to boot Phoenix/lineage/Bliss/Androidx86i only
can wating the iso.
圚 2017幎8月23日星期䞉 UTC+8䞋午7:47:37lambdadroid写道

Post by l***@gmail.com
Hi,
I personally don't use Android x86, but I maintain an Android port
for a x86 device using a Linux 4.9 kernel just like Android x86. Recently I
had to backport a few commits from AOSP's master branch to fix issues with
the newer kernel together with some Android components and I thought I'd
suggest them here if you haven't applied them already.
Note: I didn't test them on Android x86 but I assume it's affected
by the same issues as my device.
*Patch for ART crash in Linux 4.9.34+:*
There was a change in Linux 4.9.34 (and probably newer kernels too)
that causes a crash in ART when booting the device. It's related to some
memory management change that was back-ported to the 4.9 kernel. The commit
below fixes it for me.
ART: Change main-thread thread paging schemehttps://android.googlesource.com/platform/art/+/6c63f863f9dbf734732ff837c1e772b3482e59ff
*Patch for system_netd issues:*
I've had a short glance at the netd hack patch that was posted here
a few times and it looks related, but for me back-porting two commits fixed
it properly. It's related to a change that was made when upstreaming a
Set both legacy and new UID routing attributes.
https://android.googlesource.com/platform/system/netd/+/2b078678aafceeefea6a70e96ab8ddefe515d027z
Use new-style UID routing.https://android.googlesource.com/platform/system/netd/+/22c24ebd283d6e8fb782b34975417308839d4ad7
On the Android 4.9 kernel from Google, the netd issue wasn't as
obvious because it only happens when cleaning up a network interface (e.g.
when switching to VPN) with a really weird error. That's due to a kernel
bug that has been only fixed in newer kernel versions. After I back-ported
the patch below there was an error when initializing the network interface
(that was hidden before). This is fixed by the two patches I posted above.
fib_rules: fix error return codehttps://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git/commit/?id=adeb45cbb5057731ce9c47aad93756135d7947bf
I hope they are helpful for you. :)
Lambdadroid

[youling 257]

“OpenVPN for Android. I've also seen other VPN apps that are not working
properly”

it is switch vpn problem.
when boot enter Androidx86select one vpnvpn can workingcan bypass China
GFW if switch other vpnvpn not workswitch back the first select vpnalso
not working.

if you boot enter Androidselect other vpn can workingif switch
openvpnnot working.
mean is only the first select vpn can workingit is switch vpn problem.

圚 2017幎8月23日星期䞉 UTC+8䞋午8:52:52lambdadroid写道

Post by lambdadroid
I'm not sure. As I mentioned, I haven't tested this on Android x86, I use
this on a ROM that is mostly just LineageOS 14.1 with a few fixes to make
things work on a specific x86 device.
VPN worked for me with Linux 4.9 kernel, the 2 netd patches and OpenVPN
for Android. I've also seen other VPN apps that are not working properly,
though.

Post by youling 257
yes4.13 kernel vpn not working.
the Phoenix os 2.2.0 fixed vpn problemmay be them used
https://android.googlesource.com/platform/system/netd/+/22c24ebd283d6e8fb782b34975417308839d4ad7
https://android.googlesource.com/platform/system/netd/+/2b078678aafceeefea6a70e96ab8ddefe515d027
圚 2017幎8月23日星期䞉 UTC+8䞋午8:38:36lambdadroid写道

Post by lambdadroid
The kernel patch doesn't do anything except for making the problem
visible, it won't make VPN working.
For me, VPN was working after I applied the two netd patches I posted
above. It's possible that it's also working using the netd hack but I
haven't tested that.

if 4.9 kernel add this patch
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git/commit/?id=adeb45cbb5057731ce9c47aad93756135d7947bf
nougat vpn should be working ?
i haven't test vpn on 4.12/4.13 kernelhere has a netd hack for
4.12/4.13 kernel
圚 2017幎8月23日星期䞉 UTC+8䞋午8:18:39lambdadroid写道

Post by lambdadroid
Reverting the kernel patches that have caused the issue only avoids
the problem, it doesn't solve it. The patches improve kernel hardening, so
reverting them is not the best solution in my opinion.
It's an issue that should be fixed in ART, not in the kernel.

Buti from kernel source to solved art crash
https://github.com/youling257/android-4.9/commits/update?author=youling257
https://github.com/youling257/android-4.9/commits/master?after=149295eb63a3a58e458156cb1125f09d1fe86981+104
revert "mm... "i won't to build 50GB+ Android sourcei build kernel
sourceuse custom kernel to boot Phoenix/lineage/Bliss/Androidx86i only
can wating the iso.
圚 2017幎8月23日星期䞉 UTC+8䞋午7:47:37lambdadroid写道

Post by l***@gmail.com
Hi,
I personally don't use Android x86, but I maintain an Android port
for a x86 device using a Linux 4.9 kernel just like Android x86. Recently I
had to backport a few commits from AOSP's master branch to fix issues with
the newer kernel together with some Android components and I thought I'd
suggest them here if you haven't applied them already.
Note: I didn't test them on Android x86 but I assume it's affected
by the same issues as my device.
*Patch for ART crash in Linux 4.9.34+:*
There was a change in Linux 4.9.34 (and probably newer kernels too)
that causes a crash in ART when booting the device. It's related to some
memory management change that was back-ported to the 4.9 kernel. The commit
below fixes it for me.
ART: Change main-thread thread paging schemehttps://android.googlesource.com/platform/art/+/6c63f863f9dbf734732ff837c1e772b3482e59ff
*Patch for system_netd issues:*
I've had a short glance at the netd hack patch that was posted here
a few times and it looks related, but for me back-porting two commits fixed
it properly. It's related to a change that was made when upstreaming a
Set both legacy and new UID routing attributes.
https://android.googlesource.com/platform/system/netd/+/2b078678aafceeefea6a70e96ab8ddefe515d027z
Use new-style UID routing.https://android.googlesource.com/platform/system/netd/+/22c24ebd283d6e8fb782b34975417308839d4ad7
On the Android 4.9 kernel from Google, the netd issue wasn't as
obvious because it only happens when cleaning up a network interface (e.g.
when switching to VPN) with a really weird error. That's due to a kernel
bug that has been only fixed in newer kernel versions. After I back-ported
the patch below there was an error when initializing the network interface
(that was hidden before). This is fixed by the two patches I posted above.
fib_rules: fix error return codehttps://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git/commit/?id=adeb45cbb5057731ce9c47aad93756135d7947bf
I hope they are helpful for you. :)
Lambdadroid

[youling 257]

“OpenVPN for Android. I've also seen other VPN apps that are not working
properly”

it is switch to other vpn problem.
when boot enter Androidx86select one vpnvpn can workingcan bypass China
GFW if switch to other vpnvpn not workswitch back the first select
vpnalso not working.

if you boot enter Androidselect other vpn can workingif switch to
openvpnnot working.
mean is only the first select vpn can workingit is switch to others vpn
problem.

圚 2017幎8月23日星期䞉 UTC+8䞋午8:52:52lambdadroid写道

Post by lambdadroid
I'm not sure. As I mentioned, I haven't tested this on Android x86, I use
this on a ROM that is mostly just LineageOS 14.1 with a few fixes to make
things work on a specific x86 device.
VPN worked for me with Linux 4.9 kernel, the 2 netd patches and OpenVPN
for Android. I've also seen other VPN apps that are not working properly,
though.

Post by youling 257
yes4.13 kernel vpn not working.
the Phoenix os 2.2.0 fixed vpn problemmay be them used
https://android.googlesource.com/platform/system/netd/+/22c24ebd283d6e8fb782b34975417308839d4ad7
https://android.googlesource.com/platform/system/netd/+/2b078678aafceeefea6a70e96ab8ddefe515d027
圚 2017幎8月23日星期䞉 UTC+8䞋午8:38:36lambdadroid写道

Post by lambdadroid
The kernel patch doesn't do anything except for making the problem
visible, it won't make VPN working.
For me, VPN was working after I applied the two netd patches I posted
above. It's possible that it's also working using the netd hack but I
haven't tested that.

if 4.9 kernel add this patch
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git/commit/?id=adeb45cbb5057731ce9c47aad93756135d7947bf
nougat vpn should be working ?
i haven't test vpn on 4.12/4.13 kernelhere has a netd hack for
4.12/4.13 kernel
圚 2017幎8月23日星期䞉 UTC+8䞋午8:18:39lambdadroid写道

Post by lambdadroid
Reverting the kernel patches that have caused the issue only avoids
the problem, it doesn't solve it. The patches improve kernel hardening, so
reverting them is not the best solution in my opinion.
It's an issue that should be fixed in ART, not in the kernel.

Buti from kernel source to solved art crash
https://github.com/youling257/android-4.9/commits/update?author=youling257
https://github.com/youling257/android-4.9/commits/master?after=149295eb63a3a58e458156cb1125f09d1fe86981+104
revert "mm... "i won't to build 50GB+ Android sourcei build kernel
sourceuse custom kernel to boot Phoenix/lineage/Bliss/Androidx86i only
can wating the iso.
圚 2017幎8月23日星期䞉 UTC+8䞋午7:47:37lambdadroid写道

Post by l***@gmail.com
Hi,
I personally don't use Android x86, but I maintain an Android port
for a x86 device using a Linux 4.9 kernel just like Android x86. Recently I
had to backport a few commits from AOSP's master branch to fix issues with
the newer kernel together with some Android components and I thought I'd
suggest them here if you haven't applied them already.
Note: I didn't test them on Android x86 but I assume it's affected
by the same issues as my device.
*Patch for ART crash in Linux 4.9.34+:*
There was a change in Linux 4.9.34 (and probably newer kernels too)
that causes a crash in ART when booting the device. It's related to some
memory management change that was back-ported to the 4.9 kernel. The commit
below fixes it for me.
ART: Change main-thread thread paging schemehttps://android.googlesource.com/platform/art/+/6c63f863f9dbf734732ff837c1e772b3482e59ff
*Patch for system_netd issues:*
I've had a short glance at the netd hack patch that was posted here
a few times and it looks related, but for me back-porting two commits fixed
it properly. It's related to a change that was made when upstreaming a
Set both legacy and new UID routing attributes.
https://android.googlesource.com/platform/system/netd/+/2b078678aafceeefea6a70e96ab8ddefe515d027z
Use new-style UID routing.https://android.googlesource.com/platform/system/netd/+/22c24ebd283d6e8fb782b34975417308839d4ad7
On the Android 4.9 kernel from Google, the netd issue wasn't as
obvious because it only happens when cleaning up a network interface (e.g.
when switching to VPN) with a really weird error. That's due to a kernel
bug that has been only fixed in newer kernel versions. After I back-ported
the patch below there was an error when initializing the network interface
(that was hidden before). This is fixed by the two patches I posted above.
fib_rules: fix error return codehttps://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git/commit/?id=adeb45cbb5057731ce9c47aad93756135d7947bf
I hope they are helpful for you. :)
Lambdadroid

[JJ Meijer]

Thanks for the hard work hunting these patches down lambdadroid. Currently
testing after merging Google's 4.9.44 with above patches and it's looking
very good. Now only to make the Kernel stop panicking when suspending with
Intel SST driver loaded... :)

[Povilas Staniulis]

Thanks for the patches. I did workaround the netd issue by just ignoring
return value of modifyOutputInterfaceRules. Now we have a proper patch
for netd issue instead of my hack.
Out of curiosity, what device do you maintain ? Something Bay Trail
based or older ?

Post by l***@gmail.com
Hi,
I personally don't use Android x86, but I maintain an Android port for
a x86 device using a Linux 4.9 kernel just like Android x86. Recently
I had to backport a few commits from AOSP's master branch to fix
issues with the newer kernel together with some Android components and
I thought I'd suggest them here if you haven't applied them already.
Note: I didn't test them on Android x86 but I assume it's affected by
the same issues as my device.
*Patch for ART crash in Linux 4.9.34+:*
There was a change in Linux 4.9.34 (and probably newer kernels too)
that causes a crash in ART when booting the device. It's related to
some memory management change that was back-ported to the 4.9 kernel.
The commit below fixes it for me.
ART: Change main-thread thread paging scheme
https://android.googlesource.com/platform/art/+/6c63f863f9dbf734732ff837c1e772b3482e59ff
*Patch for system_netd issues:*
I've had a short glance at the netd hack patch that was posted here a
few times and it looks related, but for me back-porting two commits
fixed it properly. It's related to a change that was made when
upstreaming a change to some network routing code, which requires a
Set both legacy and new UID routing attributes.https://android.googlesource.com/platform/system/netd/+/2b078678aafceeefea6a70e96ab8ddefe515d027z
Use new-style UID routing.
https://android.googlesource.com/platform/system/netd/+/22c24ebd283d6e8fb782b34975417308839d4ad7
On the Android 4.9 kernel from Google, the netd issue wasn't as
obvious because it only happens when cleaning up a network interface
(e.g. when switching to VPN) with a really weird error. That's due to
a kernel bug that has been only fixed in newer kernel versions. After
I back-ported the patch below there was an error when initializing the
network interface (that was hidden before). This is fixed by the two
patches I posted above.
fib_rules: fix error return code
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git/commit/?id=adeb45cbb5057731ce9c47aad93756135d7947bf
I hope they are helpful for you. :)
Lambdadroid
--
You received this message because you are subscribed to the Google
Groups "Android-x86" group.
To unsubscribe from this group and stop receiving emails from it, send
Visit this group at https://groups.google.com/group/android-x86.
For more options, visit https://groups.google.com/d/optout.

[lambdadroid]

Yes, it's an ASUS Android tablet (ASUS MeMo Pad 7 ME176C) that is based on
Bay Trail. I basically threw away everything that came originally with the
device and re-made it with Linux 4.9 and components taken from Android-IA.
I use some smaller patches from Android x86 but for most things the
Android-IA components (e.g. Audio HAL or hardware accelerated video
decoder) have worked better for me, especially because some of them are
essentially just updated versions of the old closed-source components that
came originally with the device.

It's available on XDA[1] and the source is available on GitHub[2] if you
want to take a closer look.

[1]:
https://forum.xda-developers.com/memo-pad-7/orig-development/rom-unofficial-lineageos-14-1-asus-memo-t3633341
[2]: https://github.com/me176c-dev

Post by Povilas Staniulis
Thanks for the patches. I did workaround the netd issue by just ignoring
return value of modifyOutputInterfaceRules. Now we have a proper patch
for netd issue instead of my hack.
Out of curiosity, what device do you maintain ? Something Bay Trail
based or older ?

Post by l***@gmail.com
Hi,
I personally don't use Android x86, but I maintain an Android port for
a x86 device using a Linux 4.9 kernel just like Android x86. Recently
I had to backport a few commits from AOSP's master branch to fix
issues with the newer kernel together with some Android components and
I thought I'd suggest them here if you haven't applied them already.
Note: I didn't test them on Android x86 but I assume it's affected by
the same issues as my device.
*Patch for ART crash in Linux 4.9.34+:*
There was a change in Linux 4.9.34 (and probably newer kernels too)
that causes a crash in ART when booting the device. It's related to
some memory management change that was back-ported to the 4.9 kernel.
The commit below fixes it for me.
ART: Change main-thread thread paging scheme

https://android.googlesource.com/platform/art/+/6c63f863f9dbf734732ff837c1e772b3482e59ff

Post by l***@gmail.com
*Patch for system_netd issues:*
I've had a short glance at the netd hack patch that was posted here a
few times and it looks related, but for me back-porting two commits
fixed it properly. It's related to a change that was made when
upstreaming a change to some network routing code, which requires a
Set both legacy and new UID routing attributes.

https://android.googlesource.com/platform/system/netd/+/2b078678aafceeefea6a70e96ab8ddefe515d027z

Post by l***@gmail.com
Use new-style UID routing.

https://android.googlesource.com/platform/system/netd/+/22c24ebd283d6e8fb782b34975417308839d4ad7

Post by l***@gmail.com
On the Android 4.9 kernel from Google, the netd issue wasn't as
obvious because it only happens when cleaning up a network interface
(e.g. when switching to VPN) with a really weird error. That's due to
a kernel bug that has been only fixed in newer kernel versions. After
I back-ported the patch below there was an error when initializing the
network interface (that was hidden before). This is fixed by the two
patches I posted above.
fib_rules: fix error return code

https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git/commit/?id=adeb45cbb5057731ce9c47aad93756135d7947bf

Post by l***@gmail.com
I hope they are helpful for you. :)
Lambdadroid
--
You received this message because you are subscribed to the Google
Groups "Android-x86" group.
To unsubscribe from this group and stop receiving emails from it, send

<javascript:>

Post by l***@gmail.com
Visit this group at https://groups.google.com/group/android-x86.
For more options, visit https://groups.google.com/d/optout.

[youling 257]

extcon: Import Crystal Cove PMIC Power Source detection driver
extcon: Port Intel Crystal Cove power source to new intel_soc_pmic

is it can suspend to memory ?

圚 2017幎8月24日星期四 UTC+8䞊午2:33:40lambdadroid写道

Post by lambdadroid
Yes, it's an ASUS Android tablet (ASUS MeMo Pad 7 ME176C) that is based on
Bay Trail. I basically threw away everything that came originally with the
device and re-made it with Linux 4.9 and components taken from Android-IA.
I use some smaller patches from Android x86 but for most things the
Android-IA components (e.g. Audio HAL or hardware accelerated video
decoder) have worked better for me, especially because some of them are
essentially just updated versions of the old closed-source components that
came originally with the device.
It's available on XDA[1] and the source is available on GitHub[2] if you
want to take a closer look.
https://forum.xda-developers.com/memo-pad-7/orig-development/rom-unofficial-lineageos-14-1-asus-memo-t3633341
[2]: https://github.com/me176c-dev

Post by Povilas Staniulis
Thanks for the patches. I did workaround the netd issue by just ignoring
return value of modifyOutputInterfaceRules. Now we have a proper patch
for netd issue instead of my hack.
Out of curiosity, what device do you maintain ? Something Bay Trail
based or older ?

Post by l***@gmail.com
Hi,
I personally don't use Android x86, but I maintain an Android port for
a x86 device using a Linux 4.9 kernel just like Android x86. Recently
I had to backport a few commits from AOSP's master branch to fix
issues with the newer kernel together with some Android components and
I thought I'd suggest them here if you haven't applied them already.
Note: I didn't test them on Android x86 but I assume it's affected by
the same issues as my device.
*Patch for ART crash in Linux 4.9.34+:*
There was a change in Linux 4.9.34 (and probably newer kernels too)
that causes a crash in ART when booting the device. It's related to
some memory management change that was back-ported to the 4.9 kernel.
The commit below fixes it for me.
ART: Change main-thread thread paging scheme

https://android.googlesource.com/platform/art/+/6c63f863f9dbf734732ff837c1e772b3482e59ff

Post by l***@gmail.com
*Patch for system_netd issues:*
I've had a short glance at the netd hack patch that was posted here a
few times and it looks related, but for me back-porting two commits
fixed it properly. It's related to a change that was made when
upstreaming a change to some network routing code, which requires a
Set both legacy and new UID routing attributes.

https://android.googlesource.com/platform/system/netd/+/2b078678aafceeefea6a70e96ab8ddefe515d027z

Post by l***@gmail.com
Use new-style UID routing.

https://android.googlesource.com/platform/system/netd/+/22c24ebd283d6e8fb782b34975417308839d4ad7

Post by l***@gmail.com
On the Android 4.9 kernel from Google, the netd issue wasn't as
obvious because it only happens when cleaning up a network interface
(e.g. when switching to VPN) with a really weird error. That's due to
a kernel bug that has been only fixed in newer kernel versions. After
I back-ported the patch below there was an error when initializing the
network interface (that was hidden before). This is fixed by the two
patches I posted above.
fib_rules: fix error return code

https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git/commit/?id=adeb45cbb5057731ce9c47aad93756135d7947bf

Post by l***@gmail.com
I hope they are helpful for you. :)
Lambdadroid
--
You received this message because you are subscribed to the Google
Groups "Android-x86" group.
To unsubscribe from this group and stop receiving emails from it, send
Visit this group at https://groups.google.com/group/android-x86.
For more options, visit https://groups.google.com/d/optout.

[lambdadroid]

These patches are not related to suspend. They're used to detect if a
powered USB cable is connected (e.g. if the tablet is either connected to a
PC or a charger adapter).

Post by youling 257
extcon: Import Crystal Cove PMIC Power Source detection driver
extcon: Port Intel Crystal Cove power source to new intel_soc_pmic
is it can suspend to memory ?
圚 2017幎8月24日星期四 UTC+8䞊午2:33:40lambdadroid写道

Post by lambdadroid
Yes, it's an ASUS Android tablet (ASUS MeMo Pad 7 ME176C) that is based
on Bay Trail. I basically threw away everything that came originally with
the device and re-made it with Linux 4.9 and components taken from
Android-IA. I use some smaller patches from Android x86 but for most things
the Android-IA components (e.g. Audio HAL or hardware accelerated video
decoder) have worked better for me, especially because some of them are
essentially just updated versions of the old closed-source components that
came originally with the device.
It's available on XDA[1] and the source is available on GitHub[2] if you
want to take a closer look.
https://forum.xda-developers.com/memo-pad-7/orig-development/rom-unofficial-lineageos-14-1-asus-memo-t3633341
[2]: https://github.com/me176c-dev

Post by Povilas Staniulis
Thanks for the patches. I did workaround the netd issue by just ignoring
return value of modifyOutputInterfaceRules. Now we have a proper patch
for netd issue instead of my hack.
Out of curiosity, what device do you maintain ? Something Bay Trail
based or older ?

Post by l***@gmail.com
Hi,
I personally don't use Android x86, but I maintain an Android port for
a x86 device using a Linux 4.9 kernel just like Android x86. Recently
I had to backport a few commits from AOSP's master branch to fix
issues with the newer kernel together with some Android components and
I thought I'd suggest them here if you haven't applied them already.
Note: I didn't test them on Android x86 but I assume it's affected by
the same issues as my device.
*Patch for ART crash in Linux 4.9.34+:*
There was a change in Linux 4.9.34 (and probably newer kernels too)
that causes a crash in ART when booting the device. It's related to
some memory management change that was back-ported to the 4.9 kernel.
The commit below fixes it for me.
ART: Change main-thread thread paging scheme

https://android.googlesource.com/platform/art/+/6c63f863f9dbf734732ff837c1e772b3482e59ff

Post by l***@gmail.com
*Patch for system_netd issues:*
I've had a short glance at the netd hack patch that was posted here a
few times and it looks related, but for me back-porting two commits
fixed it properly. It's related to a change that was made when
upstreaming a change to some network routing code, which requires a
Set both legacy and new UID routing attributes.

https://android.googlesource.com/platform/system/netd/+/2b078678aafceeefea6a70e96ab8ddefe515d027z

Post by l***@gmail.com
Use new-style UID routing.

https://android.googlesource.com/platform/system/netd/+/22c24ebd283d6e8fb782b34975417308839d4ad7

Post by l***@gmail.com
On the Android 4.9 kernel from Google, the netd issue wasn't as
obvious because it only happens when cleaning up a network interface
(e.g. when switching to VPN) with a really weird error. That's due to
a kernel bug that has been only fixed in newer kernel versions. After
I back-ported the patch below there was an error when initializing the
network interface (that was hidden before). This is fixed by the two
patches I posted above.
fib_rules: fix error return code

https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git/commit/?id=adeb45cbb5057731ce9c47aad93756135d7947bf

Post by l***@gmail.com
I hope they are helpful for you. :)
Lambdadroid
--
You received this message because you are subscribed to the Google
Groups "Android-x86" group.
To unsubscribe from this group and stop receiving emails from it, send
Visit this group at https://groups.google.com/group/android-x86.
For more options, visit https://groups.google.com/d/optout.

[Povilas Staniulis]

Do you have camera driver working on your device ?

Post by lambdadroid
Yes, it's an ASUS Android tablet (ASUS MeMo Pad 7 ME176C) that is
based on Bay Trail. I basically threw away everything that came
originally with the device and re-made it with Linux 4.9 and
components taken from Android-IA. I use some smaller patches from
Android x86 but for most things the Android-IA components (e.g. Audio
HAL or hardware accelerated video decoder) have worked better for me,
especially because some of them are essentially just updated versions
of the old closed-source components that came originally with the device.
It's available on XDA[1] and the source is available on GitHub[2] if
you want to take a closer look.
https://forum.xda-developers.com/memo-pad-7/orig-development/rom-unofficial-lineageos-14-1-asus-memo-t3633341
[2]: https://github.com/me176c-dev
Thanks for the patches. I did workaround the netd issue by just ignoring
return value of modifyOutputInterfaceRules. Now we have a proper patch
for netd issue instead of my hack.
Out of curiosity, what device do you maintain ? Something Bay Trail
based or older ?

Post by l***@gmail.com
Hi,
I personally don't use Android x86, but I maintain an Android

port for

Post by l***@gmail.com
a x86 device using a Linux 4.9 kernel just like Android x86.

Recently

Post by l***@gmail.com
I had to backport a few commits from AOSP's master branch to fix
issues with the newer kernel together with some Android

components and

Post by l***@gmail.com
I thought I'd suggest them here if you haven't applied them

already.

Post by l***@gmail.com
Note: I didn't test them on Android x86 but I assume it's

affected by

Post by l***@gmail.com
the same issues as my device.
*Patch for ART crash in Linux 4.9.34+:*
There was a change in Linux 4.9.34 (and probably newer kernels too)
that causes a crash in ART when booting the device. It's related to
some memory management change that was back-ported to the 4.9

kernel.

Post by l***@gmail.com
The commit below fixes it for me.
ART: Change main-thread thread paging scheme

https://android.googlesource.com/platform/art/+/6c63f863f9dbf734732ff837c1e772b3482e59ff
<https://android.googlesource.com/platform/art/+/6c63f863f9dbf734732ff837c1e772b3482e59ff>

Post by l***@gmail.com
*Patch for system_netd issues:*
I've had a short glance at the netd hack patch that was posted

here a

Post by l***@gmail.com
few times and it looks related, but for me back-porting two commits
fixed it properly. It's related to a change that was made when
upstreaming a change to some network routing code, which requires a
Set both legacy and new UID routing

attributes.https://android.googlesource.com/platform/system/netd/+/2b078678aafceeefea6a70e96ab8ddefe515d027z
<https://android.googlesource.com/platform/system/netd/+/2b078678aafceeefea6a70e96ab8ddefe515d027z>

Post by l***@gmail.com
Use new-style UID routing.

https://android.googlesource.com/platform/system/netd/+/22c24ebd283d6e8fb782b34975417308839d4ad7
<https://android.googlesource.com/platform/system/netd/+/22c24ebd283d6e8fb782b34975417308839d4ad7>

Post by l***@gmail.com
On the Android 4.9 kernel from Google, the netd issue wasn't as
obvious because it only happens when cleaning up a network

interface

Post by l***@gmail.com
(e.g. when switching to VPN) with a really weird error. That's

due to

Post by l***@gmail.com
a kernel bug that has been only fixed in newer kernel versions.

After

Post by l***@gmail.com
I back-ported the patch below there was an error when

initializing the

Post by l***@gmail.com
network interface (that was hidden before). This is fixed by the

two

Post by l***@gmail.com
patches I posted above.
fib_rules: fix error return code

https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git/commit/?id=adeb45cbb5057731ce9c47aad93756135d7947bf
<https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git/commit/?id=adeb45cbb5057731ce9c47aad93756135d7947bf>

Post by l***@gmail.com
I hope they are helpful for you. :)
Lambdadroid
--
You received this message because you are subscribed to the Google
Groups "Android-x86" group.
To unsubscribe from this group and stop receiving emails from

it, send
<javascript:>

Post by l***@gmail.com
Visit this group at https://groups.google.com/group/android-x86

<https://groups.google.com/group/android-x86>.

Post by l***@gmail.com
For more options, visit https://groups.google.com/d/optout

<https://groups.google.com/d/optout>.
--
You received this message because you are subscribed to the Google
Groups "Android-x86" group.
To unsubscribe from this group and stop receiving emails from it, send
Visit this group at https://groups.google.com/group/android-x86.
For more options, visit https://groups.google.com/d/optout.

[lambdadroid]

The camera isn't working but I have no intention to make it work either.
The camera is kind of stupid on this device anyway, I doubt anyone uses it.

Post by Povilas Staniulis
Do you have camera driver working on your device ?

Post by lambdadroid
Yes, it's an ASUS Android tablet (ASUS MeMo Pad 7 ME176C) that is
based on Bay Trail. I basically threw away everything that came
originally with the device and re-made it with Linux 4.9 and
components taken from Android-IA. I use some smaller patches from
Android x86 but for most things the Android-IA components (e.g. Audio
HAL or hardware accelerated video decoder) have worked better for me,
especially because some of them are essentially just updated versions
of the old closed-source components that came originally with the

device.

Post by lambdadroid
It's available on XDA[1] and the source is available on GitHub[2] if
you want to take a closer look.

https://forum.xda-developers.com/memo-pad-7/orig-development/rom-unofficial-lineageos-14-1-asus-memo-t3633341

Post by lambdadroid
[2]: https://github.com/me176c-dev
Thanks for the patches. I did workaround the netd issue by just ignoring
return value of modifyOutputInterfaceRules. Now we have a proper patch
for netd issue instead of my hack.
Out of curiosity, what device do you maintain ? Something Bay Trail
based or older ?

Post by l***@gmail.com
Hi,
I personally don't use Android x86, but I maintain an Android

port for

Post by l***@gmail.com
a x86 device using a Linux 4.9 kernel just like Android x86.

Recently

Post by l***@gmail.com
I had to backport a few commits from AOSP's master branch to fix
issues with the newer kernel together with some Android

components and

Post by l***@gmail.com
I thought I'd suggest them here if you haven't applied them

already.

Post by l***@gmail.com
Note: I didn't test them on Android x86 but I assume it's

affected by

Post by l***@gmail.com
the same issues as my device.
*Patch for ART crash in Linux 4.9.34+:*
There was a change in Linux 4.9.34 (and probably newer kernels

too)

Post by lambdadroid

Post by l***@gmail.com
that causes a crash in ART when booting the device. It's related

to

Post by lambdadroid

Post by l***@gmail.com
some memory management change that was back-ported to the 4.9

kernel.

Post by l***@gmail.com
The commit below fixes it for me.
ART: Change main-thread thread paging scheme

https://android.googlesource.com/platform/art/+/6c63f863f9dbf734732ff837c1e772b3482e59ff

Post by lambdadroid
<

https://android.googlesource.com/platform/art/+/6c63f863f9dbf734732ff837c1e772b3482e59ff>

Post by lambdadroid

Post by l***@gmail.com
*Patch for system_netd issues:*
I've had a short glance at the netd hack patch that was posted

here a

Post by l***@gmail.com
few times and it looks related, but for me back-porting two

commits

Post by lambdadroid

Post by l***@gmail.com
fixed it properly. It's related to a change that was made when
upstreaming a change to some network routing code, which requires

a

Post by lambdadroid

Post by l***@gmail.com
Set both legacy and new UID routing

attributes.

https://android.googlesource.com/platform/system/netd/+/2b078678aafceeefea6a70e96ab8ddefe515d027z

Post by lambdadroid
<

https://android.googlesource.com/platform/system/netd/+/2b078678aafceeefea6a70e96ab8ddefe515d027z>

Post by lambdadroid

Post by l***@gmail.com
Use new-style UID routing.

https://android.googlesource.com/platform/system/netd/+/22c24ebd283d6e8fb782b34975417308839d4ad7

Post by lambdadroid
<

https://android.googlesource.com/platform/system/netd/+/22c24ebd283d6e8fb782b34975417308839d4ad7>

Post by lambdadroid

Post by l***@gmail.com
On the Android 4.9 kernel from Google, the netd issue wasn't as
obvious because it only happens when cleaning up a network

interface

Post by l***@gmail.com
(e.g. when switching to VPN) with a really weird error. That's

due to

Post by l***@gmail.com
a kernel bug that has been only fixed in newer kernel versions.

After

Post by l***@gmail.com
I back-ported the patch below there was an error when

initializing the

Post by l***@gmail.com
network interface (that was hidden before). This is fixed by the

two

Post by l***@gmail.com
patches I posted above.
fib_rules: fix error return code

https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git/commit/?id=adeb45cbb5057731ce9c47aad93756135d7947bf

Post by lambdadroid
<

https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git/commit/?id=adeb45cbb5057731ce9c47aad93756135d7947bf>

Post by lambdadroid

Post by l***@gmail.com
I hope they are helpful for you. :)
Lambdadroid
--
You received this message because you are subscribed to the Google
Groups "Android-x86" group.
To unsubscribe from this group and stop receiving emails from

it, send

<javascript:>>.

Post by lambdadroid
<javascript:>

Post by l***@gmail.com
Visit this group at https://groups.google.com/group/android-x86

<https://groups.google.com/group/android-x86>.

Post by l***@gmail.com
For more options, visit https://groups.google.com/d/optout

<https://groups.google.com/d/optout>.
--
You received this message because you are subscribed to the Google
Groups "Android-x86" group.
To unsubscribe from this group and stop receiving emails from it, send

<javascript:>

Post by lambdadroid
Visit this group at https://groups.google.com/group/android-x86.
For more options, visit https://groups.google.com/d/optout.

[Mauro Rossi]

Post by lambdadroid
Yes, it's an ASUS Android tablet (ASUS MeMo Pad 7 ME176C) that is based on
Bay Trail. I basically threw away everything that came originally with the
device and re-made it with Linux 4.9 and components taken from Android-IA.
I use some smaller patches from Android x86 but for most things the
Android-IA components (e.g. Audio HAL or hardware accelerated video
decoder) have worked better for me, especially because some of them are
essentially just updated versions of the old closed-source components that
came originally with the device.
It's available on XDA[1] and the source is available on GitHub[2] if you
want to take a closer look.
https://forum.xda-developers.com/memo-pad-7/orig-development/rom-unofficial-lineageos-14-1-asus-memo-t3633341
[2]: https://github.com/me176c-dev

I've seen you have enabled vulkan forking intel/vulkan-hal

The difference in your branch is the use of vulkan.mesa as module name
reflected in device_asus_me176 project:

https://github.com/me176c-dev/device_asus_me176c/commit/d000a057046636f2891ca637305b79d357bf4234

Are there other changes needed in init.rc or init.sh necessary to enable
the use of vulkan.intel?
Thanks for your information

Mauro

Post by lambdadroid

Post by Povilas Staniulis
Thanks for the patches. I did workaround the netd issue by just ignoring
return value of modifyOutputInterfaceRules. Now we have a proper patch
for netd issue instead of my hack.
Out of curiosity, what device do you maintain ? Something Bay Trail
based or older ?

Post by l***@gmail.com
Hi,
I personally don't use Android x86, but I maintain an Android port for
a x86 device using a Linux 4.9 kernel just like Android x86. Recently
I had to backport a few commits from AOSP's master branch to fix
issues with the newer kernel together with some Android components and
I thought I'd suggest them here if you haven't applied them already.
Note: I didn't test them on Android x86 but I assume it's affected by
the same issues as my device.
*Patch for ART crash in Linux 4.9.34+:*
There was a change in Linux 4.9.34 (and probably newer kernels too)
that causes a crash in ART when booting the device. It's related to
some memory management change that was back-ported to the 4.9 kernel.
The commit below fixes it for me.
ART: Change main-thread thread paging scheme

https://android.googlesource.com/platform/art/+/6c63f863f9dbf734732ff837c1e772b3482e59ff

Post by l***@gmail.com
*Patch for system_netd issues:*
I've had a short glance at the netd hack patch that was posted here a
few times and it looks related, but for me back-porting two commits
fixed it properly. It's related to a change that was made when
upstreaming a change to some network routing code, which requires a
Set both legacy and new UID routing attributes.

https://android.googlesource.com/platform/system/netd/+/2b078678aafceeefea6a70e96ab8ddefe515d027z

Post by l***@gmail.com
Use new-style UID routing.

https://android.googlesource.com/platform/system/netd/+/22c24ebd283d6e8fb782b34975417308839d4ad7

Post by l***@gmail.com
On the Android 4.9 kernel from Google, the netd issue wasn't as
obvious because it only happens when cleaning up a network interface
(e.g. when switching to VPN) with a really weird error. That's due to
a kernel bug that has been only fixed in newer kernel versions. After
I back-ported the patch below there was an error when initializing the
network interface (that was hidden before). This is fixed by the two
patches I posted above.
fib_rules: fix error return code

https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git/commit/?id=adeb45cbb5057731ce9c47aad93756135d7947bf

Post by l***@gmail.com
I hope they are helpful for you. :)
Lambdadroid
--
You received this message because you are subscribed to the Google
Groups "Android-x86" group.
To unsubscribe from this group and stop receiving emails from it, send
Visit this group at https://groups.google.com/group/android-x86.
For more options, visit https://groups.google.com/d/optout.

[lambdadroid]

Vulkan is indeed working for me. The name change doesn't matter, I was just
annoyed that all modules were named .baytrail when they're not specifically
related to Baytrail.

When I started the ROM I used drm_gralloc from Android x86. However, Vulkan
was not working correctly with it. It seems like the Vulkan implementation
doesn't support the GEM names that are still used in drm_gralloc. I
recently switched to Android-IA's fork of minigbm[1] which is using PRIME
fds, and since then it is working fine.

[1]: https://github.com/intel/minigbm

Post by Mauro Rossi

Post by lambdadroid
Yes, it's an ASUS Android tablet (ASUS MeMo Pad 7 ME176C) that is based
on Bay Trail. I basically threw away everything that came originally with
the device and re-made it with Linux 4.9 and components taken from
Android-IA. I use some smaller patches from Android x86 but for most things
the Android-IA components (e.g. Audio HAL or hardware accelerated video
decoder) have worked better for me, especially because some of them are
essentially just updated versions of the old closed-source components that
came originally with the device.
It's available on XDA[1] and the source is available on GitHub[2] if you
want to take a closer look.
https://forum.xda-developers.com/memo-pad-7/orig-development/rom-unofficial-lineageos-14-1-asus-memo-t3633341
[2]: https://github.com/me176c-dev

I've seen you have enabled vulkan forking intel/vulkan-hal
The difference in your branch is the use of vulkan.mesa as module name
https://github.com/me176c-dev/device_asus_me176c/commit/d000a057046636f2891ca637305b79d357bf4234
Are there other changes needed in init.rc or init.sh necessary to enable
the use of vulkan.intel?
Thanks for your information
Mauro

Post by lambdadroid

Post by Povilas Staniulis
Thanks for the patches. I did workaround the netd issue by just ignoring
return value of modifyOutputInterfaceRules. Now we have a proper patch
for netd issue instead of my hack.
Out of curiosity, what device do you maintain ? Something Bay Trail
based or older ?

Post by l***@gmail.com
Hi,
I personally don't use Android x86, but I maintain an Android port for
a x86 device using a Linux 4.9 kernel just like Android x86. Recently
I had to backport a few commits from AOSP's master branch to fix
issues with the newer kernel together with some Android components and
I thought I'd suggest them here if you haven't applied them already.
Note: I didn't test them on Android x86 but I assume it's affected by
the same issues as my device.
*Patch for ART crash in Linux 4.9.34+:*
There was a change in Linux 4.9.34 (and probably newer kernels too)
that causes a crash in ART when booting the device. It's related to
some memory management change that was back-ported to the 4.9 kernel.
The commit below fixes it for me.
ART: Change main-thread thread paging scheme

https://android.googlesource.com/platform/art/+/6c63f863f9dbf734732ff837c1e772b3482e59ff

Post by l***@gmail.com
*Patch for system_netd issues:*
I've had a short glance at the netd hack patch that was posted here a
few times and it looks related, but for me back-porting two commits
fixed it properly. It's related to a change that was made when
upstreaming a change to some network routing code, which requires a
Set both legacy and new UID routing attributes.

https://android.googlesource.com/platform/system/netd/+/2b078678aafceeefea6a70e96ab8ddefe515d027z

Post by l***@gmail.com
Use new-style UID routing.

https://android.googlesource.com/platform/system/netd/+/22c24ebd283d6e8fb782b34975417308839d4ad7

Post by l***@gmail.com
On the Android 4.9 kernel from Google, the netd issue wasn't as
obvious because it only happens when cleaning up a network interface
(e.g. when switching to VPN) with a really weird error. That's due to
a kernel bug that has been only fixed in newer kernel versions. After
I back-ported the patch below there was an error when initializing the
network interface (that was hidden before). This is fixed by the two
patches I posted above.
fib_rules: fix error return code

https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git/commit/?id=adeb45cbb5057731ce9c47aad93756135d7947bf

Post by l***@gmail.com
I hope they are helpful for you. :)
Lambdadroid
--
You received this message because you are subscribed to the Google
Groups "Android-x86" group.
To unsubscribe from this group and stop receiving emails from it, send
Visit this group at https://groups.google.com/group/android-x86.
For more options, visit https://groups.google.com/d/optout.

[Jose Luis s]

Hi Lambdadroid.

¿do you have GPS working?

Thanks

Post by lambdadroid
Yes, it's an ASUS Android tablet (ASUS MeMo Pad 7 ME176C) that is based on
Bay Trail. I basically threw away everything that came originally with the
device and re-made it with Linux 4.9 and components taken from Android-IA.
I use some smaller patches from Android x86 but for most things the
Android-IA components (e.g. Audio HAL or hardware accelerated video
decoder) have worked better for me, especially because some of them are
essentially just updated versions of the old closed-source components that
came originally with the device.
It's available on XDA[1] and the source is available on GitHub[2] if you
want to take a closer look.
https://forum.xda-developers.com/memo-pad-7/orig-development/rom-unofficial-lineageos-14-1-asus-memo-t3633341
[2]: https://github.com/me176c-dev

Post by Povilas Staniulis
Thanks for the patches. I did workaround the netd issue by just ignoring
return value of modifyOutputInterfaceRules. Now we have a proper patch
for netd issue instead of my hack.
Out of curiosity, what device do you maintain ? Something Bay Trail
based or older ?

Post by l***@gmail.com
Hi,
I personally don't use Android x86, but I maintain an Android port for
a x86 device using a Linux 4.9 kernel just like Android x86. Recently
I had to backport a few commits from AOSP's master branch to fix
issues with the newer kernel together with some Android components and
I thought I'd suggest them here if you haven't applied them already.
Note: I didn't test them on Android x86 but I assume it's affected by
the same issues as my device.
*Patch for ART crash in Linux 4.9.34+:*
There was a change in Linux 4.9.34 (and probably newer kernels too)
that causes a crash in ART when booting the device. It's related to
some memory management change that was back-ported to the 4.9 kernel.
The commit below fixes it for me.
ART: Change main-thread thread paging scheme

https://android.googlesource.com/platform/art/+/6c63f863f9dbf734732ff837c1e772b3482e59ff

Post by l***@gmail.com
*Patch for system_netd issues:*
I've had a short glance at the netd hack patch that was posted here a
few times and it looks related, but for me back-porting two commits
fixed it properly. It's related to a change that was made when
upstreaming a change to some network routing code, which requires a
Set both legacy and new UID routing attributes.

https://android.googlesource.com/platform/system/netd/+/2b078678aafceeefea6a70e96ab8ddefe515d027z

Post by l***@gmail.com
Use new-style UID routing.

https://android.googlesource.com/platform/system/netd/+/22c24ebd283d6e8fb782b34975417308839d4ad7

Post by l***@gmail.com
On the Android 4.9 kernel from Google, the netd issue wasn't as
obvious because it only happens when cleaning up a network interface
(e.g. when switching to VPN) with a really weird error. That's due to
a kernel bug that has been only fixed in newer kernel versions. After
I back-ported the patch below there was an error when initializing the
network interface (that was hidden before). This is fixed by the two
patches I posted above.
fib_rules: fix error return code

https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git/commit/?id=adeb45cbb5057731ce9c47aad93756135d7947bf

Post by l***@gmail.com
I hope they are helpful for you. :)
Lambdadroid
--
You received this message because you are subscribed to the Google
Groups "Android-x86" group.
To unsubscribe from this group and stop receiving emails from it, send
Visit this group at https://groups.google.com/group/android-x86.
For more options, visit https://groups.google.com/d/optout.

[lambdadroid]

No, GPS isn't working but to be honest I have never used it on this device
anyway... Sorry :)

Post by Jose Luis s
Hi Lambdadroid.
¿do you have GPS working?
Thanks

Post by lambdadroid
Yes, it's an ASUS Android tablet (ASUS MeMo Pad 7 ME176C) that is based
on Bay Trail. I basically threw away everything that came originally with
the device and re-made it with Linux 4.9 and components taken from
Android-IA. I use some smaller patches from Android x86 but for most things
the Android-IA components (e.g. Audio HAL or hardware accelerated video
decoder) have worked better for me, especially because some of them are
essentially just updated versions of the old closed-source components that
came originally with the device.
It's available on XDA[1] and the source is available on GitHub[2] if you
want to take a closer look.
https://forum.xda-developers.com/memo-pad-7/orig-development/rom-unofficial-lineageos-14-1-asus-memo-t3633341
[2]: https://github.com/me176c-dev

Post by Povilas Staniulis
Thanks for the patches. I did workaround the netd issue by just ignoring
return value of modifyOutputInterfaceRules. Now we have a proper patch
for netd issue instead of my hack.
Out of curiosity, what device do you maintain ? Something Bay Trail
based or older ?

Post by l***@gmail.com
Hi,
I personally don't use Android x86, but I maintain an Android port for
a x86 device using a Linux 4.9 kernel just like Android x86. Recently
I had to backport a few commits from AOSP's master branch to fix
issues with the newer kernel together with some Android components and
I thought I'd suggest them here if you haven't applied them already.
Note: I didn't test them on Android x86 but I assume it's affected by
the same issues as my device.
*Patch for ART crash in Linux 4.9.34+:*
There was a change in Linux 4.9.34 (and probably newer kernels too)
that causes a crash in ART when booting the device. It's related to
some memory management change that was back-ported to the 4.9 kernel.
The commit below fixes it for me.
ART: Change main-thread thread paging scheme

https://android.googlesource.com/platform/art/+/6c63f863f9dbf734732ff837c1e772b3482e59ff

Post by l***@gmail.com
*Patch for system_netd issues:*
I've had a short glance at the netd hack patch that was posted here a
few times and it looks related, but for me back-porting two commits
fixed it properly. It's related to a change that was made when
upstreaming a change to some network routing code, which requires a
Set both legacy and new UID routing attributes.

https://android.googlesource.com/platform/system/netd/+/2b078678aafceeefea6a70e96ab8ddefe515d027z

Post by l***@gmail.com
Use new-style UID routing.

https://android.googlesource.com/platform/system/netd/+/22c24ebd283d6e8fb782b34975417308839d4ad7

Post by l***@gmail.com
On the Android 4.9 kernel from Google, the netd issue wasn't as
obvious because it only happens when cleaning up a network interface
(e.g. when switching to VPN) with a really weird error. That's due to
a kernel bug that has been only fixed in newer kernel versions. After
I back-ported the patch below there was an error when initializing the
network interface (that was hidden before). This is fixed by the two
patches I posted above.
fib_rules: fix error return code

https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git/commit/?id=adeb45cbb5057731ce9c47aad93756135d7947bf

Post by l***@gmail.com
I hope they are helpful for you. :)
Lambdadroid
--
You received this message because you are subscribed to the Google
Groups "Android-x86" group.
To unsubscribe from this group and stop receiving emails from it, send
Visit this group at https://groups.google.com/group/android-x86.
For more options, visit https://groups.google.com/d/optout.

['dada' via Android-x86]

Hi,
Thank you provide this. but this link is broken.

Set both legacy and new UID routing attributes.
https://android.googlesource.com/platform/system/netd/+/2b078678aafceeefea6a70e96ab8ddefe515d027z


I used this:

Set both legacy and new UID routing attributes.

https://android.googlesource.com/platform/system/netd/+/2b078678aafceeefea6a70e96ab8ddefe515d027

Dada
圚 2017幎8月23日星期䞉 UTC+8䞋午7:47:37lambdadroid写道

Post by l***@gmail.com
Hi,
I personally don't use Android x86, but I maintain an Android port for a
x86 device using a Linux 4.9 kernel just like Android x86. Recently I had
to backport a few commits from AOSP's master branch to fix issues with the
newer kernel together with some Android components and I thought I'd
suggest them here if you haven't applied them already.
Note: I didn't test them on Android x86 but I assume it's affected by the
same issues as my device.
*Patch for ART crash in Linux 4.9.34+:*
There was a change in Linux 4.9.34 (and probably newer kernels too) that
causes a crash in ART when booting the device. It's related to some memory
management change that was back-ported to the 4.9 kernel. The commit below
fixes it for me.
ART: Change main-thread thread paging schemehttps://android.googlesource.com/platform/art/+/6c63f863f9dbf734732ff837c1e772b3482e59ff
*Patch for system_netd issues:*
I've had a short glance at the netd hack patch that was posted here a few
times and it looks related, but for me back-porting two commits fixed it
properly. It's related to a change that was made when upstreaming a change
Set both legacy and new UID routing attributes.
https://android.googlesource.com/platform/system/netd/+/2b078678aafceeefea6a70e96ab8ddefe515d027z
Use new-style UID routing.https://android.googlesource.com/platform/system/netd/+/22c24ebd283d6e8fb782b34975417308839d4ad7
On the Android 4.9 kernel from Google, the netd issue wasn't as obvious
because it only happens when cleaning up a network interface (e.g. when
switching to VPN) with a really weird error. That's due to a kernel bug
that has been only fixed in newer kernel versions. After I back-ported the
patch below there was an error when initializing the network interface
(that was hidden before). This is fixed by the two patches I posted above.
fib_rules: fix error return codehttps://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git/commit/?id=adeb45cbb5057731ce9c47aad93756135d7947bf
I hope they are helpful for you. :)
Lambdadroid

[youling 257]

https://android.googlesource.com/kernel/common/+/435416b698b19d2a716263804a917d3e22ed50c9
https://android.googlesource.com/kernel/common/+log/android-4.9

CONFIG_ANDROID_BINDER_IPC_SELFTEST=yboot failed on Androidx86any idea ?
please you test newest android-4.9.


<13>[ 38.864964] init: Service 'logd-reinit' (pid 3602) exited with status 0
<13>[ 38.885072] init: Starting service 'healthd'...
<13>[ 38.913354] init: Starting service 'lmkd'...
<13>[ 38.942939] init: Starting service 'servicemanager'...
<6>[ 38.943503] binder_alloc_selftest: STARTED
stopped boot .

[lambdadroid]

I don't think you're supposed to enable CONFIG_ANDROID_BINDER_IPC_SELFTEST,
that's only for testing. They add options to
kernel/configs/android-base.config or
kernel/configs/android-recommended.config if you're supposed to set them.

Anyway, I'm two commits behind right now
(at https://android.googlesource.com/kernel/common/+/8c972e1bb1865434b35cac9758c2b5a48f610393)
which is working fine for me. I'll let you know when I have tested latest.

Post by youling 257
https://android.googlesource.com/kernel/common/+/435416b698b19d2a716263804a917d3e22ed50c9
https://android.googlesource.com/kernel/common/+log/android-4.9
CONFIG_ANDROID_BINDER_IPC_SELFTEST=yboot failed on Androidx86any idea ?
please you test newest android-4.9.
<13>[ 38.864964] init: Service 'logd-reinit' (pid 3602) exited with status 0
<13>[ 38.885072] init: Starting service 'healthd'...
<13>[ 38.913354] init: Starting service 'lmkd'...
<13>[ 38.942939] init: Starting service 'servicemanager'...
<6>[ 38.943503] binder_alloc_selftest: STARTED
stopped boot .

[lambdadroid]

Just tested the latest kernel, booting fine and everything works as
expected. Try it without enabling CONFIG_ANDROID_BINDER_IPC_SELFTEST.

Post by youling 257
https://android.googlesource.com/kernel/common/+/435416b698b19d2a716263804a917d3e22ed50c9
https://android.googlesource.com/kernel/common/+log/android-4.9
CONFIG_ANDROID_BINDER_IPC_SELFTEST=yboot failed on Androidx86any idea ?
please you test newest android-4.9.
<13>[ 38.864964] init: Service 'logd-reinit' (pid 3602) exited with status 0
<13>[ 38.885072] init: Starting service 'healthd'...
<13>[ 38.913354] init: Starting service 'lmkd'...
<13>[ 38.942939] init: Starting service 'servicemanager'...
<6>[ 38.943503] binder_alloc_selftest: STARTED
stopped boot .

[youling 257]

you not help memy problem is CONFIG_ANDROID_BINDER_IPC_SELFTEST=y

i used # CONFIG_ANDROID_BINDER_IPC_SELFTEST is not setboot Androidx86 a
days.

圚 2017幎9月2日星期六 UTC+8䞋午6:37:35lambdadroid写道

Post by lambdadroid
Just tested the latest kernel, booting fine and everything works as
expected. Try it without enabling CONFIG_ANDROID_BINDER_IPC_SELFTEST.

Post by youling 257
https://android.googlesource.com/kernel/common/+/435416b698b19d2a716263804a917d3e22ed50c9
https://android.googlesource.com/kernel/common/+log/android-4.9
CONFIG_ANDROID_BINDER_IPC_SELFTEST=yboot failed on Androidx86any idea ?
please you test newest android-4.9.
<13>[ 38.864964] init: Service 'logd-reinit' (pid 3602) exited with status 0
<13>[ 38.885072] init: Starting service 'healthd'...
<13>[ 38.913354] init: Starting service 'lmkd'...
<13>[ 38.942939] init: Starting service 'servicemanager'...
<6>[ 38.943503] binder_alloc_selftest: STARTED
stopped boot .

[lambdadroid]

I'm sorry, but I'm not sure what "boot Androidx86 a days" is supposed to
mean. Could you clarify that? You shouldn't set
CONFIG_ANDROID_BINDER_IPC_SELFTEST=y.

Post by youling 257
you not help memy problem is CONFIG_ANDROID_BINDER_IPC_SELFTEST=y
i used # CONFIG_ANDROID_BINDER_IPC_SELFTEST is not setboot Androidx86 a
days.
圚 2017幎9月2日星期六 UTC+8䞋午6:37:35lambdadroid写道

Post by lambdadroid
Just tested the latest kernel, booting fine and everything works as
expected. Try it without enabling CONFIG_ANDROID_BINDER_IPC_SELFTEST.

Post by youling 257
https://android.googlesource.com/kernel/common/+/435416b698b19d2a716263804a917d3e22ed50c9
https://android.googlesource.com/kernel/common/+log/android-4.9
CONFIG_ANDROID_BINDER_IPC_SELFTEST=yboot failed on Androidx86any idea
? please you test newest android-4.9.
<13>[ 38.864964] init: Service 'logd-reinit' (pid 3602) exited with status 0
<13>[ 38.885072] init: Starting service 'healthd'...
<13>[ 38.913354] init: Starting service 'lmkd'...
<13>[ 38.942939] init: Starting service 'servicemanager'...
<6>[ 38.943503] binder_alloc_selftest: STARTED
stopped boot .

[youling 257]

you think my mean is boot time a days ? bootloops a days ?
no# CONFIG_ANDROID_BINDER_IPC_SELFTEST is not set30s boot enter
Androidx86.

圚 2017幎9月2日星期六 UTC+8䞋午7:26:06lambdadroid写道

Post by lambdadroid
I'm sorry, but I'm not sure what "boot Androidx86 a days" is supposed to
mean. Could you clarify that? You shouldn't set
CONFIG_ANDROID_BINDER_IPC_SELFTEST=y.

Post by youling 257
you not help memy problem is CONFIG_ANDROID_BINDER_IPC_SELFTEST=y
i used # CONFIG_ANDROID_BINDER_IPC_SELFTEST is not setboot Androidx86 a
days.
圚 2017幎9月2日星期六 UTC+8䞋午6:37:35lambdadroid写道

Post by lambdadroid
Just tested the latest kernel, booting fine and everything works as
expected. Try it without enabling CONFIG_ANDROID_BINDER_IPC_SELFTEST.

Post by youling 257
https://android.googlesource.com/kernel/common/+/435416b698b19d2a716263804a917d3e22ed50c9
https://android.googlesource.com/kernel/common/+log/android-4.9
CONFIG_ANDROID_BINDER_IPC_SELFTEST=yboot failed on Androidx86any idea
? please you test newest android-4.9.
<13>[ 38.864964] init: Service 'logd-reinit' (pid 3602) exited with status 0
<13>[ 38.885072] init: Starting service 'healthd'...
<13>[ 38.913354] init: Starting service 'lmkd'...
<13>[ 38.942939] init: Starting service 'servicemanager'...
<6>[ 38.943503] binder_alloc_selftest: STARTED
stopped boot .

[youling 257]

24 hours agoi known the problem is
CONFIG_ANDROID_BINDER_IPC_SELFTEST=ystopped boot Androidx86just
CONFIG_ANDROID_BINDER_IPC_SELFTEST is not setmust can booot Androidx86of
courseyesterdaymy kernel #1 can't boot enter Androidx86#2 can boot
enter Androidx86.

圚 2017幎9月2日星期六 UTC+8䞋午7:26:06lambdadroid写道

Post by lambdadroid
I'm sorry, but I'm not sure what "boot Androidx86 a days" is supposed to
mean. Could you clarify that? You shouldn't set
CONFIG_ANDROID_BINDER_IPC_SELFTEST=y.

Post by youling 257
you not help memy problem is CONFIG_ANDROID_BINDER_IPC_SELFTEST=y
i used # CONFIG_ANDROID_BINDER_IPC_SELFTEST is not setboot Androidx86 a
days.
圚 2017幎9月2日星期六 UTC+8䞋午6:37:35lambdadroid写道

Post by lambdadroid
Just tested the latest kernel, booting fine and everything works as
expected. Try it without enabling CONFIG_ANDROID_BINDER_IPC_SELFTEST.

Post by youling 257
https://android.googlesource.com/kernel/common/+/435416b698b19d2a716263804a917d3e22ed50c9
https://android.googlesource.com/kernel/common/+log/android-4.9
CONFIG_ANDROID_BINDER_IPC_SELFTEST=yboot failed on Androidx86any idea
? please you test newest android-4.9.
<13>[ 38.864964] init: Service 'logd-reinit' (pid 3602) exited with status 0
<13>[ 38.885072] init: Starting service 'healthd'...
<13>[ 38.913354] init: Starting service 'lmkd'...
<13>[ 38.942939] init: Starting service 'servicemanager'...
<6>[ 38.943503] binder_alloc_selftest: STARTED
stopped boot .

[youling 257]

your this threadnot change any thing kernel sourcepatch Android source
if CONFIG_ANDROID_BINDER_IPC_SELFTEST=yyou may know something Android
patchcan make boot enter Androidx86.

圚 2017幎9月2日星期六 UTC+8䞋午7:26:06lambdadroid写道

Post by lambdadroid
I'm sorry, but I'm not sure what "boot Androidx86 a days" is supposed to
mean. Could you clarify that? You shouldn't set
CONFIG_ANDROID_BINDER_IPC_SELFTEST=y.

Post by youling 257
you not help memy problem is CONFIG_ANDROID_BINDER_IPC_SELFTEST=y
i used # CONFIG_ANDROID_BINDER_IPC_SELFTEST is not setboot Androidx86 a
days.
圚 2017幎9月2日星期六 UTC+8䞋午6:37:35lambdadroid写道

Post by lambdadroid
Just tested the latest kernel, booting fine and everything works as
expected. Try it without enabling CONFIG_ANDROID_BINDER_IPC_SELFTEST.

Post by youling 257
https://android.googlesource.com/kernel/common/+/435416b698b19d2a716263804a917d3e22ed50c9
https://android.googlesource.com/kernel/common/+log/android-4.9
CONFIG_ANDROID_BINDER_IPC_SELFTEST=yboot failed on Androidx86any idea
? please you test newest android-4.9.
<13>[ 38.864964] init: Service 'logd-reinit' (pid 3602) exited with status 0
<13>[ 38.885072] init: Starting service 'healthd'...
<13>[ 38.913354] init: Starting service 'lmkd'...
<13>[ 38.942939] init: Starting service 'servicemanager'...
<6>[ 38.943503] binder_alloc_selftest: STARTED
stopped boot .

[lambdadroid]

I'm still confused. I understand that you'd like to find a way to boot
successfully with CONFIG_ANDROID_BINDER_IPC_SELFTEST=y (e.g. by patching
the Android sources). However, why do want to set this option? It's
supposed to be used for testing the kernel implementation, you're not
supposed to set it for real devices. If everything is working without it,
just keep it disabled.

Post by youling 257
your this threadnot change any thing kernel sourcepatch Android source
if CONFIG_ANDROID_BINDER_IPC_SELFTEST=yyou may know something Android
patchcan make boot enter Androidx86.
圚 2017幎9月2日星期六 UTC+8䞋午7:26:06lambdadroid写道

Post by lambdadroid
I'm sorry, but I'm not sure what "boot Androidx86 a days" is supposed to
mean. Could you clarify that? You shouldn't set
CONFIG_ANDROID_BINDER_IPC_SELFTEST=y.

Post by youling 257
you not help memy problem is CONFIG_ANDROID_BINDER_IPC_SELFTEST=y
i used # CONFIG_ANDROID_BINDER_IPC_SELFTEST is not setboot Androidx86 a
days.
圚 2017幎9月2日星期六 UTC+8䞋午6:37:35lambdadroid写道

Post by lambdadroid
Just tested the latest kernel, booting fine and everything works as
expected. Try it without enabling CONFIG_ANDROID_BINDER_IPC_SELFTEST.

Post by youling 257
https://android.googlesource.com/kernel/common/+/435416b698b19d2a716263804a917d3e22ed50c9
https://android.googlesource.com/kernel/common/+log/android-4.9
CONFIG_ANDROID_BINDER_IPC_SELFTEST=yboot failed on Androidx86any
idea ? please you test newest android-4.9.
<13>[ 38.864964] init: Service 'logd-reinit' (pid 3602) exited with status 0
<13>[ 38.885072] init: Starting service 'healthd'...
<13>[ 38.913354] init: Starting service 'lmkd'...
<13>[ 38.942939] init: Starting service 'servicemanager'...
<6>[ 38.943503] binder_alloc_selftest: STARTED
stopped boot .

[youling 257]

the problem is libart[ 117.904234] zygote[4701]: segfault at bff9e000 ip
b297a910 sp bffbcb70 error 4 in libart.so[b244b000+729000]

this patch
https://android.googlesource.com/platform/art/+/6c63f863f9dbf734732ff837c1e772b3482e59ff
fix libart

buti used nougat xposedthe xposed hook libarti like use xposed.
Source (for dev):
Xposed ART: https://github.com/PurifyOS/xposed_art_n
Xposed App_process: https://github.com/PurifyOS/xposed_app_process
XposedBridge: https://github.com/PurifyOS/XposedBridge
XposedTools: https://github.com/PurifyOS/XposedTools

sosomeone can build x86/x86_64 nougat xposed include this
https://android.googlesource.com/platform/art/+/6c63f863f9dbf734732ff837c1e772b3482e59ff
patch ?

[youling 257]

https://github.com/PurifyOS/xposed_art_n/commits/aosp-7.1.2_r27
https://github.com/PurifyOS/xposed_art_n/commits/aosp-7.1.2_r27-noJIT
what different ?
i used 26-07-17-xposed-v87-sdk25-Xposed-by-PurifyOS.zip has a monthno
crashstablei don't know it used which xposed_art branch.
sotwo version xposed_artsomeone can build ?