解决休眠底电流过大调试步骤

Comment
Dear customer  Ok, you can try solution 00024639 , may be it's related with mipi ulps mode .  Thanks
我将LCD屏摘掉电流是可以下得去的，为2.2mA，我先检查屏的影响。谢谢！
Dear customer  make "target_cont_splash_screen" function in lk return 0 .  change cont_splash_enabled to 0 in mdp_pdata structure @ board file .  Thanks
Dear customer  please disable continue splash display .  Thanks
Dear customer  Change USE_OPENGL_RENDERER to false .  And change composition to mdp :  adb shell setprop debug.composition.type mdp  adb shell stop  adb shell start  And use follow method to get mdp and gpu clk :  adb shell mount –t debugfs none /sys/kernel/debug  adb shell cat /sys/kernel/debug/clk/grp_3d_clk/rate ( grp_3d_p_clk )  adb shell cat /sys/kernel/debug/clk/mdp_clk/rate  'rate' is the clock value as requested by the software. 'rate' should be a stable value when in suspend state. please get it, and we can compare with sample phone.  Thanks
Dear customer  下面是我另外提得case的回复：  I am not meaning to get the value of kgsl-3d0 interrupts count once, but check its value regularly (ex, 1 time every second) when enter sleep.  If its value dose not increase during sleep, then we are sure the 3d application is turely sleeping.  Thanks  我按照他所说的，在休眠稳定后马上插上USB查看kgsl-3d0的值是1131，然后再拔掉USB进休眠，再插上USB查看kgsl-3d0的值是1133，基本上没有变化，（屏正常显示的时候kgsl-3d0值变化很快），那就是kgsl-3d0已经休下去了。
Through your two log, my analysis result as:  1.From dmesg_3.txt, it looks there is a modem task will wake up the cpu.  2.From dmesg_4.txt, it looks GRP_3D_P_CLK is not close.I doubt there is a 3d application works and don't sleep.  My suggestions as:  First, you'd better submit a case to display team to check "GRP_3D_P_CLK is not close" and confirm the phone can sleep in the fight mode.
dmesg_3.txt 是在普通模式下的log  dmesg_4.txt 是在飞行模式下的log
请查看log
Dear customer:  Please confirm your phone works in the flight mode.  Thanks!  2013/04/09
Dear customer:  It looks GRP_3D_P_CLK is not close.It is a 3D graphical's clock.  Please check wether or not there is a 3d application work and it don't sleep.
dear customer:  echo 1 >/sys/module/pm2/parameters/debug_mask  To enable wakeup IRQ's debug mask  echo 1 > /sys/module/msm_show_resume_irq/parameters/debug_mask  kmesg.
我打开了那个clock——name，请查看附件
Dear customer:  Through the call, I know your question is how to follow "task smdcntl0 keep the lock", but before that please do a test as:  Please modify the code in the clock_debug_print_enabled() from $(AP)/kernel/arch/arm/mach-msm/clock-debug.c as the following way.  change "// pr_info("\t%s\n", clk->dbg_name);" to "pr_info("\t%s\n", clk->dbg_name);"  Then re-build boot.img by "make bootimage" and upload
dear customer:  <6>[ 1812.919621] [(1980-01-06 01:37:46.110281703 UTC)] [cpuid: 0] Enabled clock count: 18  <6>[ 1812.919859] [(1980-01-06 01:37:46.110523370 UTC)] [cpuid: 0] wakeup wake lock: smdcntl0  task smdcntl0 持锁，请检查这个任务。  谢谢
附件是我打的dmesg的log，adb shell里面不能使用grep，不好意思。您将就着看。
Please do the following test:  在执行suspend动作之前先 echo 1> /data/debug/clk/debug_suspend,之后拔下USB，等系统睡眠后再联上usb，通过 adb shell dmesg | grep 'Enabled clock' 看看clk的情况。  The following info is an example normal sleep log in the 7x27A platform and you can give me your log.  "  <6>[ 234.097167] AHB_M_CLK  <6>[ 528.405976] AHB_S_CLK <6>[ 234.097167] AHB_S_CLK  <6>[ 528.405976] CSI0_CLK <6>[ 234.097167] CSI0_CLK  <6>[ 528.405976] CSI1_CLK <6>[ 234.097167] CSI1_CLK  <6>[ 528.405976] dsi_esc_clk <6>[ 234.097167] dsi_esc_clk  <6>[ 528.405976] DSI_REF_CLK <6>[ 234.097167] DSI_REF_CLK  <6>[ 528.405976] EBI1_CLK <6>[ 234.097167] EBI1_CLK  <6>[ 528.405976] EBI2_CLK <6>[ 234.097167] EBI2_CLK  <6>[ 528.405976] MDP_DSI_P_CLK <6>[ 234.097167] MDP_DSI_P_CLK  <6>[ 528.405976] PBUS_CLK <6>[ 234.097167] PBUS_CLK  <6>[ 528.405976] PCM_CLK <6>[ 234.097167] PCM_CLK  <6>[ 528.405976] SDC2_CLK <6>[ 234.097167] ebi_acpu_clk  <6>[ 528.405976] SDC2_P_CLK <6>[ 234.097167] Enabled clock count: 12  "
没加过clock
Dear customer:  MODM: 20136317 SLEEP: MOD_ENTER_TCXO: 000013b9 00000000 00000000 //睡眠的时间很短,正常睡眠的机器，这个值会大很多，是：06ddcdb5  MODM: 20141286 SLEEP: MOD_TCXO_END: 00000000 00000001 00000000 //唤醒的原因是timetick interrupt,就是说系统因为定时器到时而产生中断唤醒机  Do you add some new clock in the modem?  Thanks!
Dear customer:  I'm sorry, there is no process temporarily.If there is some good news, I'll connect you at once.
按照您说的，把检测CPU温度的进程关掉了，log也没显示，但测出来的底电流跟刚开始一样！电流没有变化。
Dear customer:  init.qcom.post_boot.sh is in the $(AP)/device/qcom/common/rootdir/etc/*.It looks it is in the system.img.  You need exec "make systemimage" to build new system.img.  Thanks!
Dear customer:  The ThermalDaemon will detect the temperature of the cpu, if the temperature is too high the frequency of the cpu will be reduced.  I think the detect speed is too high. You can test to stop it by cancelling "start thermald" in the init.qcom.post_boot.sh.
附件传hal层的log，请查收。  E/ThermalDaemon( 887): Sensor 'MSM_THERM' - 21*C (21504)  上面那句话好像休眠的时候都在打印，每隔大概半秒左右！
按照你最后的要求，重新上传！
上传dump_sym 和 dump_voters
Dear customer:  There is the new info.  First, please do the following work before you get the smemlog and voters_log:  1.adb shell insmod /system/lib/modules/sleep_monitor.ok;  2.adb shell echo 3 > /sys/class/sleep_monitor/sleepm/control.  Then, when you collect the smem log, please do "adb shell cat /mnt/debugfs/smem_log/dump_sym_cont > sleepdebug.txt" for a 2 mins after you have got the smemlog and voters_log.  Last, upload smemlog, voters_log and sleepdebug.txt to me.
Dear customer:  We can know the bp is not sleep because there are something against sleep through  "MODM: 30498218 SLEEP: NO_SLEEP_NEW: 00000000 00000000 00000001  MODM: 30705106 SLEEP: NO_SLEEP_NEW: 00000000 00000000 00004005  MODM: 30705803 SLEEP: NO_SLEEP_NEW: 00000000 00000000 00000001" in the rpm log.  Please upload the smemlog and voters_log again as:  "adb pull /data/debug/smem_log/dump_sym;adb pull /data/debug/smem_log/dump_voters".
按照电话讲的，我将内核和share mem的log重新打了一下。
Sorry customer:  I can't get more info from your smem log in the attachment.There is a way to get the more accurate log as the following info.  1 先连接TRACE，用attach让手机跑起来。  2.按power键，让手机进入sleep状态，等待5s，按power键让系统起来。  (目的是可以看来系统sleep状态是否正常)  3. 起来后立刻break位，用下面方法将smem log导出来。  (1). do smemlog.cmm脚本抓取smem log  (2). 打开MS-DOS  (3). 将MS-DOS 中CD 到xxxx\modem_proc\tools\debug目录  (4). 在已生成.lst文件目录下执行  smem_log.pl > smem.log  会将生成的.lst文件生成smem.log文件
补上上传的smem log。
现在我烧了我们最新的软件版本，换上最新的QCN，底电流是15mA左右，内核log附件。共享内存的log我再研究下，晚些时候传上来。
Dear customer:  You can know what's the smem log refer to "80-VR692-1_C_SMEM_Log_User_Guide".  Thanks!
Dear customer:  You can get share memory as the following info.  #mkdir /data/debug  #mount -t debugfs debugfs /data/debug  The /data/debug/smem_log/dump_sym is the share memory log.  Thanks!
dear customer:  1.检查QCN是否正确烧写  2.检查硬件电路  3.检查AP侧是否睡眠（内核log）  4.检查modem侧是否睡眠（共享内存log）  谢谢

转自：http://blog.csdn.net/dwyane_zhang/article/details/8786407