[参考译文] Linux：无法加载 PVRSRVKM/SGX 驱动程序

https://e2e.ti.com/support/processors-group/processors/f/processors-forum/615731/linux-pvrsrvkm-sgx-driver-fails-to-load

在 Linux 论坛上发布了我的查询，没有得到任何答复：

我们有一个基于 Jacinto 6的定制板，通过 Linux 4.45启动它时附带以下 SDK

ti-processor-sdk-linux-automotive-dra7xx-evm-03.02.00.03-installer.bin

加载 GPU 驱动程序'pvrsrvkm.ko"时遇到问题

#insmod pvrsrvkm.ko

[46.464932]  powerdomain：GPU_pwrdm：等待时间过长、无法完成转换
[46.634790]  powerdomain：GPU_pwrdm：等待时间过长、无法完成转换
[46.804621]  powerdomain：GPU_pwrdm：等待时间过长、无法完成转换
[46.814292]  OMAP-hwmod：GPU：_wait_target_ready 失败：-16
[46.982827]  powerdomain：GPU_pwrdm：等待时间过长、无法完成转换
[46.992798]  PVR_K：(错误)：EnableSGXClocks：PM_Runtime_get_sync 失败(16)
[47.000075]  PVR_K：(错误)：SysInitialize：无法启用系统时钟(175)

它看起来是 GPU Powerdomain 问题、在插入驱动程序时未解决。

为了确认这一点、我们显示了以下调试消息：

diff --git a/arch/arm/mach-omap2/prm44xx.c b/arch/arm/mach-omap2/prm44xx.c
索引8f26621..39a28ba 100644
-- a/arch/arm/mach-omap2/prm44xx.c
++ b/arch/arm/mach-omap2/prm44xx.c
@@-422、6 +422、8 @@静态空 omap44xx_PRM_CLEAR_LOSS_FLAGS_OLD (u8器件、S16安装、
 
 静态 int omar4_pwrdm_set_next pwrst (struct powerdomain * pwrdm、u8 pwrst)
 ｛
+      pr_alert ("OMAP4_pwrdm_set_next _pwrst 为%d\n"、pwrst)；
+       
       OMAP4_prminst_RMW_inst_reg_bits (OMAP_powerstate_mask、
                                       (pwrst << omap_powerstate_shift)、
                                       pwrdm->prcm_partition、
@@-650、8 +652、29 @@静态 int omAP4_pwrdm_wait_transition (struct powerdomain * pwrdm)
              (C++< PWRdm_transition _援助))
               udelay (1)；
 
+
/*      打开状态*/
+        if (！strcmp (pwrdm->name、"GPU_pwrdm"))
+      ｛
+              pr_info ("******* Insie OMAP4_PM_PWSTCTRL_WRITE ******* \n");
+              OMAP4_prminst_RMW_AP4_REG_Bits (0x3、3、pwrdm->prcm_partition、pwrdm->prcm_fs、OMinst_PM_PWSTCTRL)；
+      ｝
+
+      if (！strcmp (pwrdm->name、"GPU_pwrdm"))
+      ｛
+              pr_alert (" READ_OMAP4_PM_PWSTCTRL 的值为%x\n"、OMAP4_prminst_Read_inst_reg (pwrdm->prcm_partition、pwrdm->prcm_fs、
+                                       OMAP4_PM_PWSTCTRL)；
+      ｝
+
+        if (！strcmp (pwrdm->name、"GPU_pwrdm"))
+      ｛
+              pr_alert ("**" PM_PW_STCTRL_SGX 的值 is_prm4x %x\n"、OMAP4_prminst_read_inst_reg (pwrdm->prcm_partition、pwrdm->prcm_fs、OMAP4_PM_PWSTST)；
+             /* pr_alert ("**PM_PW_STCTRL_SGX 的值为_1_prm4x %x\n"、 readl_relaxed (pwrdm->prcm_fs + OMAP4_PM_PWSTST))；*/
+               
+      ｝
+
       if (c > PWRDM_TRANSION_EVAOUT)｛
-              pr_err ("powerdomain：%s：等待时间过长、无法完成转换\n"、
+              pr_err ("powerdomain！！！！！！！！ ：%s：等待时间过长，无法完成转换\n",
                      pwrdm->name)；
               RETURN -EAGAIN；
       ｝

OMAP4_PM_PWSTST 读取"0x100000"、这意味着第20位被置位、并且正在进行电源域处理。尽管输入、我们可以将 OMAP4_PM_PWSTST 的值读取为0x37。

OMAP4_PM_PWSTST 和 OMAP4_PM_PWSTCTRL 的值以 dmesg 的两个不同间隔捕获

[0.000000]     正常区域：内存映射使用1536页
[0.000000]     正常区域：保留0页
[0.000000]     正常区域：196608页、LIFO 批处理：31
[0.000000]     HIGHMEM 区域：327167页、LIFO 批处理：31
[0.000000]   DRA752 ES1.1
[0.000000]   检查电源域是否始终处于打开状态***** 30000
[0.0000]*********    Insie OMAP4_PM_PWSTCTRL_WRITE *******
[0.000000]   READ_OMAP4_PM_PWSTCTRL 的值为30003
[0.000000]   PM_PW_STCTRL_SGX 的值为_prm4x 0
[0.000000]   OMAP4_pwrdm_set_next _pwrst 为3
[0.000000]   OMAP4_pwrdm_set_next _pwrst 为3
[0.000000]   OMAP4_pwrdm_set_next _pwrst 为3
[0.000000]   OMAP4_pwrdm_set_next _pwrst 为3
[0.000000]   OMAP4_pwrdm_set_next _pwrst 为3
[0.000000]   OMAP4_pwrdm_set_next _pwrst 为3
[0.000000]   OMAP4_pwrdm_set_next _pwrst 为3
[0.000000]   OMAP4_pwrdm_set_next _pwrst 为3
[0.000000]   OMAP4_pwrdm_set_next _pwrst 为3
[0.000000]   OMAP4_pwrdm_set_next _pwrst 为3
[0.000000]   OMAP4_pwrdm_set_next _pwrst 为3
[0.000000]   OMAP4_pwrdm_set_next _pwrst 为3
[0.000000]   OMAP4_pwrdm_set_next _pwrst 为3
[0.000000]   OMAP4_pwrdm_set_next _pwrst 为3
[0.000000]   OMAP4_pwrdm_set_next _pwrst 为3
[0.000000]   OMAP4_pwrdm_set_next _pwrst 为3
[0.000000]   OMAP4_pwrdm_set_next _pwrst 为3
[0.000000]   OMAP4_pwrdm_set_next _pwrst 为3
[0.000000]   OMAP4_pwrdm_set_next _pwrst 为3
[0.000000]   OMAP4_pwrdm_set_next _pwrst 为3
[0.000000]   OMAP4_pwrdm_set_next _pwrst 为3
[0.000000]   OMAP4_pwrdm_set_next _pwrst 为3
[0.000000]   OMAP4_pwrdm_set_next _pwrst 为3
[0.000000]   检查电源域是否始终处于打开状态***** 30003.
[0.0000]*********    Insie OMAP4_PM_PWSTCTRL_WRITE *******
[0.000000]   READ_OMAP4_PM_PWSTCTRL 的值为30003
[0.000000]   PM_PW_STCTRL_SGX is _prm4x 37的值
[0.000000]   检查电源域是否始终处于打开状态***** 30003.
[0.0000]*********    Insie OMAP4_PM_PWSTCTRL_WRITE *******
[0.000000]   READ_OMAP4_PM_PWSTCTRL 的值为30003
[0.000000]   PM_PW_STCTRL_SGX is _prm4x 37的值
[0.000000]   PERCPU：嵌入式12页/CPU @eef34000 s19584 r8192 d21376 u49152
[0.000000]   pcpu-alloc：s19584 r8192 d21376 u49152 alloc=12*4096
[0.000000]   pcpu-alloc：[0][1] 1.

[2.823839]   TI OMAP4 +器件的电源管理。
[2.828743]   OMAP4_pwrdm_set_next _pwrst 为0
[2.832974]   OMAP4_pwrdm_set_next _pwrst 为0
[2.837177]   OMAP4_pwrdm_set_next _pwrst 为0
[2.841404]   OMAP4_pwrdm_set_next _pwrst 为0
[2.845608]   OMAP4_pwrdm_set_next _pwrst 为0
[2.849835]   OMAP4_pwrdm_set_next _pwrst 为0
[2.854038]   OMAP4_pwrdm_set_next _pwrst 为0
[2.858239]   OMAP4_pwrdm_set_next _pwrst 为0
[2.862462]   OMAP4_pwrdm_set_next _pwrst 为1
[2.866664]   OMAP4_pwrdm_set_next _pwrst 为0
[2.870895]   OMAP4_pwrdm_set_next _pwrst 为3
[2.875098]   OMAP4_pwrdm_set_next _pwrst 为0
[2.879302]   OMAP4_pwrdm_set_next _pwrst 为0
[2.883523]   OMAP4_pwrdm_set_next _pwrst 为0
[2.887725]   OMAP4_pwrdm_set_next _pwrst 为0
[2.891953]   OMAP4_pwrdm_set_next _pwrst 为0
[2.896171]   OMAP4_pwrdm_set_next _pwrst 为3
[2.900402]   OMAP4_pwrdm_set_next _pwrst 为3
[2.904648]   检查电源域是否始终打开***** 30000
[2.910155    Insie OMAP4_PM_PWSTCTRL_WRITE *******
[2.915750]   READ_OMAP4_PM_PWSTCTRL 的值为30003
[2.920734]   PM_PW_STCTRL_SGX is _prm4x 0的四个值
[2.925720]   检查电源域是否始终打开***** 30003.
[2.931226    Insie OMAP4_PM_PWSTCTRL_WRITE *******
[2.936821]   READ_OMAP4_PM_PWSTCTRL 的值为30003
[2.941804]   PM_PW_STCTRL_SGX 是_prm4x 0的四个值

有人能告诉我们，为什么在插入驱动程序时电源域关闭？