I am posting here because I do not have permission to post in the android dev forum. Thanks to j00m who helped post my ROM earlier in the dev forum, but I thought I should build up my posts to be able to post stuff myself too Here is another smaller system.img file that should work with stock hboot or blackrose (not tested) and has the home button working. The ROM should be considered alpha quality at best as is with other CM9 ROMs.
(system.img)
http://www.mediafire.com/?9lcz59qq8s1qgvb
boot.img (if you have one from drewis' alphas, you do not need to flash this.)
http://www.mediafire.com/?kilz59rjc8kug1m
Be sure to wipe data/cache before reboot. You should be able to use the google apps from userdata.img from one of drewis's alphas.
The ROM is built from CM9, with my changes to make use of the adreno GL drivers. Credits are due to the CM team and all whose code is included in CM9. There should be many to name and I do not want to miss out anyone who deserves credit.
Will post code changes soon. Need to clean it up. Hence the delay.
I hope that when I have enough posts to be privileged, this thread can be moved into the dev forum.
Enjoy!!
Here is the source patch.
diff --git a/frameworks/base/libs/gui/SurfaceTexture.cpp b/frameworks/base/libs/gui/SurfaceTexture.cpp
index c72a45b..9e425e3 100644
--- a/libs/gui/SurfaceTexture.cpp
+++ b/libs/gui/SurfaceTexture.cpp
@@ -246,6 +246,8 @@ status_t SurfaceTexture::requestBuffer(int slot, sp<GraphicBuffer>* buf) {
return NO_ERROR;
}
+#define MAX_SLEEPTIMEOUTS (5)
+
status_t SurfaceTexture::dequeueBuffer(int *outBuf, uint32_t w, uint32_t h,
uint32_t format, uint32_t usage) {
ST_LOGV("SurfaceTexture::dequeueBuffer");
@@ -262,6 +264,9 @@ status_t SurfaceTexture::dequeueBuffer(int *outBuf, uint32_t w, uint32_t h,
int found, foundSync;
int dequeuedCount = 0;
bool tryAgain = true;
+ int sleepTimeouts = 0;
+ bool eglWorkAround = true;
+
while (tryAgain) {
if (mAbandoned) {
ST_LOGE("dequeueBuffer: SurfaceTexture has been abandoned!");
@@ -342,6 +347,21 @@ status_t SurfaceTexture::dequeueBuffer(int *outBuf, uint32_t w, uint32_t h,
// clients are not allowed to dequeue more than one buffer
// if they didn't set a buffer count.
if (!mClientBufferCount && dequeuedCount) {
+ LOGV("SurfaceTexture::dequeue: Not allowed to dequeue more than a "
+ "buffer\n");
+ if (eglWorkAround) {
+ if (sleepTimeouts++ < MAX_SLEEPTIMEOUTS) {
+ LOGD("SurfaceTexture::dequeue: Not allowed to dequeue more "
+ "than a buffer SLEEPING\n");
+ usleep(100000);
+ } else {
+ mClientBufferCount = mServerBufferCount;
+ LOGD("SurfaceTexture::dequeue: Not allowed to dequeue more "
+ "than a buffer RETRY mBufferCount:%d mServerBufferCount:%d\n",
+ mBufferCount, mServerBufferCount);
+ }
+ continue;
+ }
return -EINVAL;
}
@@ -353,6 +373,13 @@ status_t SurfaceTexture::dequeueBuffer(int *outBuf, uint32_t w, uint32_t h,
// than allowed.
const int avail = mBufferCount - (dequeuedCount+1);
if (avail < (MIN_UNDEQUEUED_BUFFERS-int(mSynchronousMode))) {
+ if (eglWorkAround && mClientBufferCount != 0) {
+ mBufferCount++;
+ mClientBufferCount = mServerBufferCount = mBufferCount;
+ LOGD("SurfaceTexture::dequeuebuffer: MIN EXCEEDED "
+ "mBuffer:%d bumped\n", mBufferCount);
+ continue;
+ }
ST_LOGE("dequeueBuffer: MIN_UNDEQUEUED_BUFFERS=%d exceeded "
"(dequeued=%d)",
MIN_UNDEQUEUED_BUFFERS-int(mSynchronousMode),
@@ -700,8 +727,8 @@ status_t SurfaceTexture::updateTexImage() {
ST_LOGW("updateTexImage: clearing GL error: %#04x", error);
}
- glBindTexture(mTexTarget, mTexName);
- glEGLImageTargetTexture2DOES(mTexTarget, (GLeglImageOES)image);
+ glBindTexture(getCurrentTextureTarget(), mTexName);
+ glEGLImageTargetTexture2DOES(getCurrentTextureTarget(), (GLeglImageOES)image);
bool failed = false;
while ((error = glGetError()) != GL_NO_ERROR) {
@@ -736,7 +763,7 @@ status_t SurfaceTexture::updateTexImage() {
mDequeueCondition.signal();
} else {
// We always bind the texture even if we don't update its contents.
- glBindTexture(mTexTarget, mTexName);
+ glBindTexture(getCurrentTextureTarget(), mTexName);
}
return OK;
@@ -762,6 +789,9 @@ bool SurfaceTexture::isExternalFormat(uint32_t format)
}
GLenum SurfaceTexture::getCurrentTextureTarget() const {
+ if (mTexTarget == GL_TEXTURE_EXTERNAL_OES) {
+ return GL_TEXTURE_2D;
+ }
return mTexTarget;
}
diff --git a/frameworks/base/services/surfaceflinger/Layer.cpp b/frameworks/base/services/surfaceflinger/Layer.cpp
index 317cc3b..9e92e81 100644
--- a/services/surfaceflinger/Layer.cpp
+++ b/services/surfaceflinger/Layer.cpp
@@ -40,6 +40,8 @@
#include "SurfaceTextureLayer.h"
#define DEBUG_RESIZE 0
+#define GL_TEXTURE_EXTERNAL_OES GL_TEXTURE_2D
+
namespace android {
diff --git a/device/htc/passion-common/BoardConfigCommon.mk b/device/htc/passion-common/BoardConfigCommon.mk
index b86c580..e17da36 100644
--- a/BoardConfigCommon.mk
+++ b/BoardConfigCommon.mk
@@ -73,7 +73,7 @@ BOARD_VENDOR_USE_AKMD := akm8973
## Hardware rendering
## doesn't actually work until we have hwcomposer
-USE_OPENGL_RENDERER := true
+USE_OPENGL_RENDERER := false
BOARD_EGL_CFG := device/htc/passion-common/egl.cfg
## for rmcc egl hack
COMMON_GLOBAL_CFLAGS += \
On Media fire:
http://www.mediafire.com/?dxzyyfzzp9m6fu8
It works with OPENGL_RENDERER := true as well, but is quite slow and crashes often, with OPENGL_RENDERER := false things are much more stable and usable as is the ROM posted above.
Thanks.
Good job guy.Keep it up ,waiting for next version
mori90 said:
Good job guy.Keep it up ,waiting for next version
Click to expand...
Click to collapse
+1 Agreed. Cant wait to see what you and texasice does next.
Confirm it is working with blackrose. But what i noticed:
-Superuser crashes when trying to update the su binary.
-My USM App is corrupted. Filesize is 12 kb, instead of 38. When trying to use it (activate UMS) it prompts a toast saying "Superuser permission denied", but it doesn't even promt the request popup before; though there is no entry in the Superuser App list.
EDIT: After rm Superuser.apk and flashing the signed Superuser zip via recovery, using su works now.
That made it possible to use the UMS App again and also to use Titanium to restore all Gapps (incl. Account Setup).
Could you check the permissions on the actual su binary itself?
rapmv78 said:
Could you check the permissions on the actual su binary itself?
Click to expand...
Click to collapse
not any longer. :/ (see edited post above) btw: did you made any changes to my ums app? noticed the smaller file size and seems like some notifications are not there any longer!?
Most likely that the permission got fixed when you flashed the Superuser zip file. Good to know it is resolved. Have you seen any other issues pertaining to graphics? I am just curious to know how GL works for others
rapmv78 said:
Have you seen any other issues pertaining to graphics? I am just curious to know how GL works for others
Click to expand...
Click to collapse
So far not. Everything is fine. Played a few videos with diefferent formats without any issues amd now i'm populating the system with apps and restoring settings.. One thing (not gl related) i'm struggeling with, is to get the market running. for whatever reason every version i installed opens fine, let me connect to my google account but then says network error (in wifi and also mobile data mode). Same error with Maps.. I'll post updates as soon as i have some. But beside this, everything looks really good. Meanwhile i can offer a system report if that helps you to get any information: http://www.nahklick.de/user/stephan/android_report.html
Bexton said:
So far not. Everything is fine. Played a few videos with diefferent formats without any issues amd now i'm populating the system with apps and restoring settings.. One thing (not gl related) i'm struggeling with, is to get the market running. for whatever reason every version i installed opens fine, let me connect to my google account but then says network error (in wifi and also mobile data mode). Same error with Maps.. I'll post updates as soon as i have some. But beside this, everything looks really good. Meanwhile i can offer a system report if that helps you to get any information: http://www.nahklick.de/user/stephan/android_report.html
Click to expand...
Click to collapse
Regarding the market, did you try clearing the cache for the app and reboot (the usual method iirc) ? System report looks fine. Thanks.
rapmv78 said:
Regarding the market, did you try clearing the cache for the app and reboot (the usual method iirc) ?
Click to expand...
Click to collapse
yep, didn't solved it!? but stopped trying now.
at the moment i'm watching spiderman 3 in 720p x264 mkv format. trailer worked fine. full movie judderes a bit from time to time but it is running for 15 minutes now without hanging up or whatever.
edit: 1080p also works but is really slow and the judder effect is much bigger.
Thread moved to dev section and link to mediafire ok
What's working and what's not?
imfloflo said:
Thread moved to dev section and link to mediafire ok
Click to expand...
Click to collapse
second post, the link to media fire is not ok!
@j00m Fix
Thread moved in General Section to allow OP to respond for the moment
This ROM enables hardware acceleration for most of the tasks as used to be in gingerbread. Am interested in knowing the experience of others using this mod.
Thanks.
Trackball wakeup will have next next... version,right ???
Why this ver. is working perfect on wifi(Static IP) but texasice version is not ??
I am not sure about the difference with static IP working in this ROM v/s texas ice's. Also since I posted the source code and is in CM9, you should just be able to get more features out of the other ROMs than you would from me This ROM was more of a proof of concept to see if openGL acceleration patch was usable and looks like it does. Thanks for your help in testing it. I do not have any plans to release new ROMs, as all I would be doing is pull the code and compiling it, without any value add from my side (now that the code has already been merged).
The mod can actually close this thread if need be.
Thanks.
what's working and what's not?
[From my README]
A mod for the Oneplus One DWC3 otg module. This allows for charging and host mode simultaneously, inspired by Ziddey's msm_otg mod for the Nexus 4/7 (2013). Functionality was ported over from his kernel hack to the DWC3 USB driver which now handles the MSM8974 USB controller.
The hack works through setting a custom module parameter I've added to allow 'ACA' host mode. This flag effectively turns on ID_A host mode while disabling VBUS power going to the hosted device. I've uploaded the modded dwc3_otg.c file that you can replace in your Oneplus One(bacon) kernel source of choice. It'll be located in the drivers/usb/dwc3/ directory. I've also uploaded my personal kernel image with this hack built on top of Franco's kernel. It also has other modules built into it, mainly DRM/Devtmpfs/Cifs/NFS/NTFS/Alsa Sequencer/Usbip/Binfmt/loadable modules/etc... It was compiled with GCC 4.9 NDK version.
Usage: First you'll need either a generic Y split USB OTG cable or a powered USB hub connected to regular OTG(I've only tested the Y cable).
With the modified kernel flashed, open a terminal shell and as root, enter the following command: "echo Y > /sys/module/dwc3/parameters/aca_enable"
This activates the 'ACA' host mode hack.
[UPDATE]
Wiggling the cord is not needed anymore. I've updated the code so that the phone automatically accepts a charge on entering host mode.
The tricky part is now getting your Y-OTG adapter to send power to the phone. First with power cable and USB device(s) connected the adapter, plug the Y cable into the phone. Test that the phone reads the device. Now, unplug the cable from the phone, leaving the USB device and power cord plugged into the OTG adapter. Gently wiggle the cable slightly while slowly pushing it back into the phone's port, wait for the charge indicator to come on. Once the phone detects the charge, you can push the cable in all the way. The phone should be charging at max current rate while retaining host mode.
Please, if you can test the powered hub method or have improvements to this hack, feel free to share! Also, the standard legal disclaimer applies here that by using this mod/code/kernel in anyway is completely your responsibility. I'm not liable for any possible damages to your devices.
Links:
DWC3 OTG Modification For OnePlus One
https://github.com/sollapse/opo_dwc3_otg/
Ziddey's Original ACA hack for Mako
https://github.com/ziddey/mako/commits/nightlies-4.3-JSS
Franco's Oneplus One Kernel Source
https://github.com/franciscofranco/one_plus_one
Hi sollapse, thank you a lot for this patch, i modified it a little for my xperia z2, it seems to be working just fine, it detects when power is disconnected and switches back to OTG_STATE_A_IDLE automatically, anyway, the only thing that i couldnt get to work is once it is on host mode, when i connect the power it doesnt start charging, perhaps you can find a way to get it working, here's my patch:
Code:
--- dwc3_otg.c 2015-09-22 14:44:45.115324661 -0300
+++ dwc3_otg.c.new 2015-09-22 14:47:40.950239287 -0300
@@ -45,6 +45,12 @@ static void dwc3_otg_reset(struct dwc3_o
static void dwc3_otg_notify_host_mode(struct usb_otg *otg, int host_mode);
static void dwc3_otg_reset(struct dwc3_otg *dotg);
+/*OTG charging hack*/
+static bool aca_enable = 0;
+static bool enable_otg_charge = 0;
+module_param(enable_otg_charge, bool, S_IRUGO | S_IWUSR);
+MODULE_PARM_DESC(enable_otg_charge, "Force ACA host mode to allow charging and host.");
+
/**
* dwc3_otg_set_host_regs - reset dwc3 otg registers to host operation.
*
@@ -204,41 +210,45 @@ static int dwc3_otg_start_host(struct us
if (!dwc->xhci)
return -EINVAL;
-
- if (!dotg->vbus_otg) {
- dotg->vbus_otg = devm_regulator_get(dwc->dev->parent,
- "vbus_dwc3");
- if (IS_ERR(dotg->vbus_otg)) {
- dev_err(dwc->dev, "Failed to get vbus regulator\n");
- ret = PTR_ERR(dotg->vbus_otg);
- dotg->vbus_otg = 0;
- return ret;
- }
- }
-
+
+ if(!aca_enable){
+ if (!dotg->vbus_otg) {
+ dotg->vbus_otg = devm_regulator_get(dwc->dev->parent,
+ "vbus_dwc3");
+ if (IS_ERR(dotg->vbus_otg)) {
+ dev_err(dwc->dev, "Failed to get vbus regulator\n");
+ ret = PTR_ERR(dotg->vbus_otg);
+ dotg->vbus_otg = 0;
+ return ret;
+ }
+ }
+ }
+
if (on) {
dev_dbg(otg->phy->dev, "%s: turn on host\n", __func__);
dwc3_otg_notify_host_mode(otg, on);
-
- /* register ocp notification */
- if (ext_xceiv && ext_xceiv->otg_capability &&
- ext_xceiv->ext_ocp_notification.notify) {
- ret = regulator_register_ocp_notification(
- dotg->vbus_otg,
- &ext_xceiv->ext_ocp_notification);
- if (ret)
- dev_err(otg->phy->dev,
- "unable to register ocp\n");
- }
-
- ret = regulator_enable(dotg->vbus_otg);
- if (ret) {
- dev_err(otg->phy->dev, "unable to enable vbus_otg\n");
- dwc3_otg_notify_host_mode(otg, 0);
- return ret;
- }
-
+
+ if(!aca_enable){
+ /* register ocp notification */
+ if (ext_xceiv && ext_xceiv->otg_capability &&
+ ext_xceiv->ext_ocp_notification.notify) {
+ ret = regulator_register_ocp_notification(
+ dotg->vbus_otg,
+ &ext_xceiv->ext_ocp_notification);
+ if (ret)
+ dev_err(otg->phy->dev,
+ "unable to register ocp\n");
+ }
+
+ ret = regulator_enable(dotg->vbus_otg);
+ if (ret) {
+ dev_err(otg->phy->dev, "unable to enable vbus_otg\n");
+ dwc3_otg_notify_host_mode(otg, 0);
+ return ret;
+ }
+ }
+
/* The delay between enabling regulator and adding the
platform device is needed to succeed in the enumeration
for certain devices. */
@@ -269,7 +279,9 @@ static int dwc3_otg_start_host(struct us
dev_err(otg->phy->dev,
"%s: failed to add XHCI pdev ret=%d\n",
__func__, ret);
- regulator_disable(dotg->vbus_otg);
+ if(!aca_enable)
+ regulator_disable(dotg->vbus_otg);
+
dwc3_otg_notify_host_mode(otg, 0);
return ret;
}
@@ -279,24 +291,27 @@ static int dwc3_otg_start_host(struct us
dwc3_otg_reset(dotg);
} else {
dev_dbg(otg->phy->dev, "%s: turn off host\n", __func__);
-
- ret = regulator_disable(dotg->vbus_otg);
- if (ret) {
- dev_err(otg->phy->dev, "unable to disable vbus_otg\n");
- return ret;
- }
-
- /* unregister ocp notification */
- if (ext_xceiv && ext_xceiv->otg_capability &&
- ext_xceiv->ext_ocp_notification.notify) {
- ret = regulator_register_ocp_notification(
- dotg->vbus_otg, NULL);
- if (ret)
- dev_err(otg->phy->dev,
- "unable to unregister ocp\n");
- }
-
- dwc3_otg_notify_host_mode(otg, on);
+
+ if(!aca_enable){
+ ret = regulator_disable(dotg->vbus_otg);
+ if (ret) {
+ dev_err(otg->phy->dev, "unable to disable vbus_otg\n");
+ return ret;
+ }
+
+ /* unregister ocp notification */
+ if (ext_xceiv && ext_xceiv->otg_capability &&
+ ext_xceiv->ext_ocp_notification.notify) {
+ ret = regulator_register_ocp_notification(
+ dotg->vbus_otg, NULL);
+ if (ret)
+ dev_err(otg->phy->dev,
+ "unable to unregister ocp\n");
+ }
+ }
+
+ if(!aca_enable)
+ dwc3_otg_notify_host_mode(otg, on);
platform_device_del(dwc->xhci);
/*
@@ -341,8 +356,11 @@ static int dwc3_otg_set_host(struct usb_
* required for XHCI controller before setting OTG Port Power
* TODO: Tune this delay
*/
- msleep(300);
- dwc3_otg_set_host_power(dotg);
+
+ msleep(300);
+ if(!aca_enable){
+ dwc3_otg_set_host_power(dotg);
+ }
} else {
otg->host = NULL;
}
@@ -972,7 +990,17 @@ static void dwc3_otg_sm_work(struct work
dev_dbg(phy->dev, "vbus_drop_det\n");
/* staying on here until exit from A-Device */
} else {
- phy->state = OTG_STATE_A_HOST;
+ //if we have power and enable_otg_charge, force charging on
+ if (test_bit(B_SESS_VLD, &dotg->inputs) && enable_otg_charge){
+ if(charger){
+ dev_info(phy->dev, "OTG charging is ON!!!\n");
+ aca_enable = 1;
+ charger->chg_type =DWC3_SDP_CHARGER;
+ dwc3_otg_set_power(phy, DWC3_IDEV_CHG_MAX);
+ }
+ } else
+ aca_enable = 0;
+ phy->state = OTG_STATE_A_HOST;
ret = dwc3_otg_start_host(&dotg->otg, 1);
if ((ret == -EPROBE_DEFER) &&
dotg->vbus_retry_count < 3) {
@@ -1000,7 +1028,15 @@ static void dwc3_otg_sm_work(struct work
break;
case OTG_STATE_A_HOST:
- if (test_bit(ID, &dotg->inputs)) {
+ if (enable_otg_charge){
+ //detect power change and switch back to OTG_STATE_A_IDLE
+ if(!test_bit(B_SESS_VLD, &dotg->inputs)){
+ dev_info(phy->dev, "Power disconnected, charging disabled!!!\n");
+ dwc3_otg_start_host(&dotg->otg, 0);
+ phy->state = OTG_STATE_A_IDLE;
+ work = 1;
+ }
+ } else if (test_bit(ID, &dotg->inputs)) {
dev_dbg(phy->dev, "id\n");
dwc3_otg_start_host(&dotg->otg, 0);
phy->state = OTG_STATE_B_IDLE;
I've updated my code a couple of days ago to automatically apply power during the state change, assuming a charger is connected to the y cable (USB devices would not power otherwise). In your mod, it seems you're attempting to detect a charge using the 'B' mode state, but by that time with the aca flag enabled, the USB controller should be in ID_A host mode instead of B_SESS_VLD. The driver's 'set_power' function already checks the line for current automatically. Please look at the updated code to see what I did during the state change. It should all work utilizing the one parameter flag.
I tested this with a patched sultanXDA CAF kernel (since the .c source file was slightly different, I was careful to only merge the relevant changes), and a powered USB HUB. It works perfectly! Thanks.
Phoenix Wright said:
I tested this with a patched sultanXDA CAF kernel (since the .c source file was slightly different, I was careful to only merge the relevant changes), and a powered USB HUB. It works perfectly! Thanks.
Click to expand...
Click to collapse
That's great! I'm pretty certain with the push to USB Type C and potential USB 3.1 support, that the DWC3 driver will be used across most devices for now on. This code should merge well with them if ACA is absent (Zenphone 2 is the only exception I've found so far since Intel enabled it in the driver).
Can someone post a modded CAF kernel for the opo? I am having issues building a kernel as I currently only have a windows box.
snekiam said:
Can someone post a modded CAF kernel for the opo? I am having issues building a kernel as I currently only have a windows box.
Click to expand...
Click to collapse
Latest Sultanxda kernel (as of 10/4), with the CAF dwc_otg modified with sollapse's patches. His sources are here: https://github.com/sultanxda/android_kernel_oneplus_msm8974
I used Google GCC 4.8, which is the same one he uses, as far as I could tell.
View attachment sultan10-4_otg-y.7z
@sollapse I noticed a bug. I made two widgets with an app on Play Store (to enable and disable this hack). I disabled this by mistake while the OTG device and charge were still inserted, and after I removed them I got a kernel panic, could this be fixed?
I should be able to fix it. I'll look at it again and will try to clean up the state change code.
Kernel
Can someone please upload a kernel .img?
I am not able to compile a kernel on my own....
@sollapse I fixed the kernel panics, and also another bug (when the aca_enable parameter was on, you couldn't charge at full speed with a wall charger - not in host mode, just simple charging). I've done all possible status changes (set aca_enable to on, insert otg-y cable + device, turn aca_enable off, unplug device; aca_enable is off, plug regular otg cable + device, turn aca_enable on, unplug device, set aca_enable to off, re-plug). These two situations respectively gave these results: kernel panic and inability to use regular otg again.
Turns out that the issues were with "regulator_disable", now it's executed when the regulator is enabled, regardless of the aca_enable parameter.
Another fix for incorrect usage: if you connected a regular OTG cable+device while aca_enable was set to Y, it would show the charging icon, and it would even stay there after the device was disconnected! Sadly I can't prevent the charging icon from appearing (as this hack basically forces all the "unknown" chargers to be reported as USB - as opposed to AC - chargers... and the weird part is that the charge coming from OTG-Y is an "unknown charger"... but the charge (?) coming from a regular OTG device is an "unknown charger" too! XD, if there's a way it's out of reach for me), but I managed to make it go away after disconnection. Basically, it doesn't force "unknown" to USB if the function gets called to *disable* charging. I guess this is the best I can do for this kernel mod ^_^
It looks I was wrong, after lots of debugging of dwc3_otg I learned how it works, so I overhauled the patch. It works more cleanly now: it actually properly detects the charger (no more phantom notifications, faster charging rates if it's allowed by the charger), supports unplugging and replugging of the power cord while using OTG-Y, and unplugging of the device after power was unplugged, fixes all sorts of race conditions and wrong usages of the aca_enable parameter (I'm not sure if it's all of them, but I've been testing and fixing things for a while).
I attach a .diff for CAF kernels. Take note that to use the fix for plugging OTG devices in sleep mode you need to patch the charger driver: https://github.com/sultanxda/androi...mmit/7b023b295fdfa4789c93aee8b04f5d2a9b52dbba
Wow, I've completely abandoned this hack since it's worked well enough for me . Also haven't had much time due to work and other necessities. Thanks for the patches @Phoenix Wright for cleaning up this hack! This should be a standard addition to all future Android kernels.
sollapse said:
Wow, I've completely abandoned this hack since it's worked well enough for me . Also haven't had much time due to work and other necessities. Thanks for the patches @Phoenix Wright for cleaning up this hack! This should be a standard addition to all future Android kernels.
Click to expand...
Click to collapse
Yeah, I wonder why they haven't implemented the feature officially, as it can be done just fine
And many thanks for developing this
By the way, I realized the reboots when inserting an OTG-Y device in deep sleep were not actually fixed, as I got a reboot yesterday, my bad (sultanxda fixed the issues with doing it with regular OTG, but it seems the issue with OTG-Y was different). It seems it's fixed properly this time though. Same patch as yesterday, just a two line difference.
Thanks for the original mod, sollapse, and thanks for your patches and all the updates, Phoenix. Just to confirm: this mod requires that on every kernel update via Sultan's OTA, the kernel source be pulled, patched, flashed onto your OPO, and finally activated using the aca command?
I'm off to go learn how to build a kernel from this handy guide, if anyone else is interested in jumping on this exciting bandwagon.
http://forum.xda-developers.com/android/software/ultimate-guide-compile-android-kernel-t2871276
EDIT: Thought I'd keep everyone abrest of my progress as a total newbie trying this out. I've used Ubunutu before, but had a lot of headaches trying to get my network working properly.
5:00pm: started the downloads of Ubuntu and Virtualbox. I have Hyper-V support (which is supposedly faster) on my desktop rig, but it has network issues, so I went with Virtualbox. Apparently, one of the things needed to create a kernel is a "toolchain". The guide uses "arm-eabi-4.33", but Phoenix and apparently sultan use gcc 4.8, so we'll go with that. But, how do I replace them? Can I just switch the git urls with this one I found of Google's toolchain?
5:30pm: got Ubuntu's 15.10 iso downloaded and VirtualBox installed. We'll figure out the toolchains later.
5:45pm: OK, Ubuntu is installing. I did the default on everything (2GB RAM, 8GB virtual hard disk). Booted up, installed it, and rebooted. "SQUASHFS errors" on first Ubuntu boot. The issues begin early, lol.
5:50pm: Checked the iso's MD5, matches up. Google'd around, but only random fixes that don't really fit (updating my BIOS? on a virtualbox?!). OK, let's try again with 3GB of RAM (my rig has 8GB) and 12GB of disk space.
5:55pm: "SquashFS errors" again. Whhhhhhhyyyy, Ubuntu, whyyyyyyyyy.
5:56pm: Hmm, what if I just shut down the virtual box and just try starting Ubuntu again?
5:57pm: HOLYCOWITWORKED. Thank you, Ubuntu--I love you.
5:58pm: Why is the resolution so small even though Ubuntu is installed? Everything is huge and my mouse is a little laggy.
6:05pm: OK, from a StackExchange post: you need "Guest Additions" to change the resolution. In the VirtualBox window (not in Ubuntu), you just click "Install Guest Additions" and it does its thing. It actually inserts the files as a virtual CD, haha. That's neat.
6:15pm: Trying to get the hang of "Right Control" as "host key". But, wait, why can't I copy/paste between my host and guest (copying all the commands from the guide linked above)? I feel like I had that last time I used VirtualBox. The Firefox in Ubunutu is really slow, too, and troubleshooting is mostly through Google right now, lol, and I don't want slow troubleshooting.
6:20pm: Oh, bidirectional copy/paste needs to be enabled. I see....
6:22pm: I FEEL LIKE A GOD NOW. I can copy and paste between TWO operating systems running my PC. WOWZA....I don't know what enromous amount of engineering that required, but it is a killer feature.
6:26pm: I'm trying to learn all the commands that the guide is using, just in case I mess up somewhere. What does the "-y" do in apt-get?
6:26:15pm: Thank you, random internet stranger many years ago, about telling me what a "man page" is. It's a simple website/HTML document that will actually explain all the arguments of a command. Great in interpreting the exotic arguments of cerebral StackExchange users....and what "-y" does in apt-get....
6:26:30pm: "-y" just hits "yes" to any prompts or whatever that apt-get invokes. Nice noob feature for people like me, haha. All right, I'm OK with that argument, let's do it!
6:30pm: Next, "mkdir"....well, that makes sense. But, where is it making this folder? How does it know where to put it????
6:33pm: Ah, so there is a "Home" folder that is the default folder. That's where it puts things. I made like 15 folders using "mkdir" and they all appear there.
6:40pm: Let's start! First, I need to grab sultan's kernel. OK, "git clone", let's do it!
6:41pm: "Git is not installed". These errors freak me out, but at least this one is minor, lol. OK: step zero, install git.
6:42pm: THE COMMAND WORKED! First success. But, wow, the kernel is a huge download (300MB+). Damn, wish I started this earlier in the background.
6:44pm: I wonder if I can start installing the other things, too. But, how? Can I open another terminal window?
6:45pm: Yes, you can open another terminal window and even a tab. But, should I install other things? Might that interfere with "git clone"? Why risk it? I mean, it took two hours to get just here...
6:46pm: Throwing caution to the wind. Opening TWO terminal windows and starting apt-get with the other stuff.
6:50pm: No errors so far...
6:55pm: woot! Everything downloaded, nice, nice. OK, the last thing: those toolchains. What even is a toolchain?
6:56pm: OK, a toolchain is basically like a list of directories and their location? Or something? It's like a default variable location "list" or something. Hmm...OK, Phoenix used gcc 4.8, so that's what we'll do!
6:58pm: But, wait, the instructions at the gcc toolchain page say I need the "Android GCC repository" and even more dependencies. But, the guide doesn't mention any of that. Is this specific to the Google one? Looks like more downloading...
6:59pm: Err, hold on. I don't want to BUILD the toolchain. I just want it cloned to my system....I don't need to build the toolchain, right? OK, let's just get the gcc repo first and then we'll see.
7:05pm: Ugh, another huge download. I'm down to 180MB of free space! What? How did I use 12GB already?
7:06pm: Oh, the HDD is only 9.6GB. I guess formatting takes away some of it and Ubuntu's install, too. OK, we'll resize the partition and call it night. To be continued later...
A brief update, as I'm too tired to annotate everything today, haha. The reason the guide uses Doom's toolchains is because they're prebuilt. So, I guess, we'll use those. I'm creating the updated dwc3_otg.c from all the patches, but the hunks are failing after the first patch. Will troubleshoot today by looking at the reject file and see what's being moved around.
EDIT: OK, I actually just opened the patch files. Looks like I just need the latest one to apply to sollapse's original patch. Patches 1/2 are almost identical (and sans full overhaul). Patch 3 is superseded by patch 4. So, just patch 4 is the one you need. Huh. Let's do it! ;D
This guide is great for understanding patch files.
sollapse said:
Wow, I've completely abandoned this hack since it's worked well enough for me . Also haven't had much time due to work and other necessities. Thanks for the patches @Phoenix Wright for cleaning up this hack! This should be a standard addition to all future Android kernels.
Click to expand...
Click to collapse
thanks for your work, it's awesome feature
is there some solution without a kernel compiling ? some custom rom or kernel image
stadnyuk said:
thanks for your work, it's awesome feature
is there some solution without a kernel compiling ? some custom rom or kernel image
Click to expand...
Click to collapse
Its already there in my kernel called lightning kernel check tht sig for the link and if you are on cm then there is another thread in original android dev section
nikhil18 said:
Its already there in my kernel called lightning kernel check tht sig for the link and if you are on cm then there is another thread in original android dev section
Click to expand...
Click to collapse
Thank you, I'll check it. I've installed sultanxda ROM with hope for future kernel updates, but there is some bugs and I think to return to stock CM
nikhil18 said:
Its already there in my kernel called lightning kernel check tht sig for the link and if you are on cm then there is another thread in original android dev section
Click to expand...
Click to collapse
Hi again, I'm stuck with lightning kernel(((
Can't flash it. I'm starting to think I'm completely noon.
Flashing over night build of CM12 the cm version of kernel(V15)
Wipe, flash, wipe
The result - stuck on one plus logo
Flashing again sultanxda ROM, wipe, kernel, gapps
The result - boot loop
Where is the problem? What I'm doing wrong?
Now I'm on a last night build, am I right that I need V15 cm version?
Thank you for helping
PS: I've realized I was flashing cm13 kernel version with cm12 ROM, so flashed sultanxda 12, v33 kernel version - loop boot
Basically, there has been an app ported to Android that allows even unrooted(stock) devices to deliver a bootrom exploit to the Nintendo Switch via USB-OTG and a USB cable (or C-to-C). USB 3.0 (xHCI) devices have no issues and deliver the exploit just fine. Apparently it is not even a USB 2.0 problem but rather how the EHCI performs, as certain USB 2.0 phones actually have the xHCI controller and can run the exploit just fine. What happens is that although it can detect the connected Switch in Tegra Recovery Mode, it just doesn't do anything and gives an error in the logs, "SUMBITURB failed".
On Linux desktop systems it is similar, but the exploit can still work with a kernel patch provided by a hacking group that discovered the exploit in the first place:
Code:
--- linux-4.14.27/drivers/usb/host/ehci-hcd.c.old 2018-04-17 18:00:00.000000000 +0000
+++ linux-4.14.27/drivers/usb/host/ehci-hcd.c 2018-04-17 18:00:00.000000000 +0000
@@ -873,14 +873,6 @@
INIT_LIST_HEAD (&qtd_list);
switch (usb_pipetype (urb->pipe)) {
- case PIPE_CONTROL:
- /* qh_completions() code doesn't handle all the fault cases
- * in multi-TD control transfers. Even 1KB is rare anyway.
- */
- if (urb->transfer_buffer_length > (16 * 1024))
- return -EMSGSIZE;
- /* FALLTHROUGH */
- /* case PIPE_BULK: */
default:
if (!qh_urb_transaction (ehci, urb, &qtd_list, mem_flags))
return -ENOMEM;
The author of the Android port had also written a Python "hotpatch" script for desktop Linux systems:
Code:
#!/usr/bin/env python3
import os
"""
Cursed Code.
This code literally patches your kernel memory, proceed at your own risk.
Tested on Ubuntu 17.10 and Arch, x86_64. Should work on other distros, maybe even other architectures!
Run fusee-launcher.py with the "--override-checks" argument.
If you'd rather patch your drivers properly:
https://github.com/fail0verflow/shofel2/blob/master/linux-ehci-enable-large-ctl-xfers.patch
"""
ksyms = {
line[2]: int(line[0], 16)
for line in
map(lambda l: l.strip().split(),
open("/proc/kallsyms", "r").readlines())}
print(hex(ksyms["ehci_urb_enqueue"]))
patch_c = """
#include <linux/module.h>
#include <linux/kernel.h>
#include <asm/pgtable.h>
static u32 ORIG_MAX = 16*1024;
static u32 NEW_MAX = 0x1000000;
/* borrowed from MUSL because I'm lazy AF */
static char *fourbyte_memmem(const unsigned char *h, size_t k, const unsigned char *n)
{
uint32_t nw = n[0]<<24 | n[1]<<16 | n[2]<<8 | n[3];
uint32_t hw = h[0]<<24 | h[1]<<16 | h[2]<<8 | h[3];
for (h+=3, k-=3; k; k--, hw = hw<<8 | *++h)
if (hw == nw) return (char *)h-3;
return 0;
}
static pte_t* (*lookup_addr)(unsigned long, unsigned int*) = (void *) PLACE2;
static void set_addr_rw(unsigned long addr) {
unsigned int level;
pte_t *pte = lookup_addr(addr, &level);
set_pte_atomic(pte, pte_mkwrite(*pte));
}
int init_module(void) {
void * ehci_urb_enqueue_start = (void *) PLACEHOLDER;
u32 * patch_addr;
printk(KERN_INFO "Patch module loaded\\n");
patch_addr = (u32 *) fourbyte_memmem(ehci_urb_enqueue_start, 0x400, (void *)&ORIG_MAX);
if (patch_addr == NULL) {
printk(KERN_INFO "Failed to find patch site :(\\n");
return -1;
}
printk(KERN_INFO "patch_addr: 0x%px\\n", patch_addr);
set_addr_rw((unsigned long)patch_addr);
*patch_addr = NEW_MAX;
printk(KERN_INFO "Patching done!\\n");
return -1;
}
""".replace("PLACEHOLDER", hex(ksyms["ehci_urb_enqueue"])).replace("PLACE2", hex(ksyms["lookup_address"]))
makefile = """
obj-m += patch.o
all:
make -C /lib/modules/$(shell uname -r)/build M=$(PWD) modules
"""
with open("patch.c", "w") as patchfile:
patchfile.write(patch_c)
with open("Makefile", "w") as mf:
mf.write(makefile)
os.system("make")
print("About to insert patch module, 'Operation not permitted' means it probably worked, check dmesg output.")
os.system("insmod patch.ko")
I tried to see if running it in Termux would do anything but I got the following error:
Code:
0x0
Traceback (most recent call last):
File "ehci_patch.py", line 70, in <module>
" " ".replace("PLACEHOLDER", hex(ksyms["ehci_urb_enqueue"])).replace("PLACE2" hex(ksyms["lookup_address"]))
KeyError: 'lookup_address'
I know that script isn't meant for use on Android anyway but maybe it can lead to a solution. The author of it does not know how to go about it at this time either, but believes an entire recompile of the kernel would be necessary. I am hoping that something like a systemless Magisk module would be the easiest solution for users but do not know if that is possible. I am only guessing it might be possible to create a Magisk module because of audio drivers like VIPER4Android. If indeed a custom kernel is needed, does anyone know how to go about it? It could be difficult to implement for everyone because not everyone has a device where the source to the kernel is available, etc. I am willing, however, to test anything on my tablet which is USB 2.0 and gives the error in the app. Any advice for how to go about this will be greatly appreciated.
I feel ya man, i need this stuff too. NXLoader doesn't work on my Galaxy Grand Prime (G530T) and i really need it to Dx