RTL8127网卡导致UEFI 出现Unhandled Exception in EL3

Robotics & Edge Computing Jetson Systems Jetson Thor
,
1 
▒▒Unhandled Exception in EL3.
x30            = 0x000000000ff5a384
x0             = 0x0000000000000000
x1             = 0x8000000000000003
x2             = 0x0000000040100000
x3             = 0x000000000000001a
x4             = 0x0000001fb639ef98
x5             = 0x0000000000000000
x6             = 0x0000000000000019
x7             = 0x0000000000000001
x8             = 0x0000000000000000
x9             = 0x000000000000001a
x10            = 0x0000001fb6361f78
x11            = 0x0000000000000000
x12            = 0x0000000000000000
x13            = 0x0000000200000000
x14            = 0x0000000000000000
x15            = 0x0000000200000000
x16            = 0x0000001fb7192630
x17            = 0x0000000000000000
x18            = 0x0000000000000000
x19            = 0x000000000000000a
x20            = 0x0000000040100010
x21            = 0x0000001fb639efa8
x22            = 0x0000000000000000
x23            = 0x0000000000000001
x24            = 0x0000000000000001
x25            = 0x0000000000020000
x26            = 0x0000000000000000
x27            = 0x0000000000000000
x28            = 0x0000000040100000
x29            = 0x0000001fd3ffe1c0
scr_el3        = 0x000000401c07073d
sctlr_el3      = 0x00000000b0cd183f
cptr_el3       = 0x0000000000000100
tcr_el3        = 0x0000000080853510
daif           = 0x00000000000003c0
mair_el3       = 0x00000000004404ff
spsr_el3       = 0x00000000420002cd
elr_el3        = 0x000000000ff5a584
ttbr0_el3      = 0x000000000ff99001
esr_el3        = 0x00000000be000011
far_el3        = 0xfa6dcf59dfaede7c
spsr_el1       = 0x0000000000000000
elr_el1        = 0x0000000000000000
spsr_abt       = 0x0000000000000000
spsr_und       = 0x0000000000000000
spsr_irq       = 0x0000000000000000
spsr_fiq       = 0x0000000000000000
sctlr_el1      = 0x0000000030d00980
actlr_el1      = 0x0000000000000000
cpacr_el1      = 0x0000000000300000
csselr_el1     = 0x0000000000000002
sp_el1         = 0x0000000000000000
esr_el1        = 0x0000000000000000
ttbr0_el1      = 0x0000000000000000
ttbr1_el1      = 0x0000000000000000
mair_el1       = 0x0000000000000000
amair_el1      = 0x0000000000000000
tcr_el1        = 0x0000000000000000
tpidr_el1      = 0x0000000000000000
tpidr_el0      = 0x0000000080000000
tpidrro_el0    = 0x0000000000000000
par_el1        = 0x0000000000000800
mpidr_el1      = 0x0000000081000000
afsr0_el1      = 0x0000000000000000
afsr1_el1      = 0x0000000000000000
contextidr_el1 = 0x0000000000000000
vbar_el1       = 0x0000000000000000
cntp_ctl_el0   = 0x0000000000000005
cntp_cval_el0  = 0x000000021c64f871
cntv_ctl_el0   = 0x0000000000000000
cntv_cval_el0  = 0x0000000000000000
cntkctl_el1    = 0x0000000000000000
sp_el0         = 0x0000001fb72ef100
isr_el1        = 0x0000000000000040
cpuectlr_el1   = 0xc0007403c0543001
gicd_ispendr regs (Offsets 0x200 - 0x278)
 Offset:                        value
0000000000000200:               0x0000000000000000
0000000000000204:               0x0000000000000000
0000000000000208:               0x0000000000000000
000000000000020c:               0x0000000000000008
0000000000000210:               0x0000000010000000
0000000000000214:               0x0000000000000000
0000000000000218:               0x0000000000000000
000000000000021c:               0x0000000000000000
0000000000000220:               0x0000000000000000
0000000000000224:               0x0000000000000000
0000000000000228:               0x0000000000000000
000000000000022c:               0x0000000000000000
0000000000000230:               0x0000000000000000
0000000000000234:               0x0000000000000000
0000000000000238:               0x0000000080000000
000000000000023c:               0x0000000000000000
0000000000000240:               0x0000000000000000
0000000000000244:               0x0000000000000000
0000000000000248:               0x0000000000000000
000000000000024c:               0x0000000000000000
0000000000000250:               0x0000000000001800
0000000000000254:               0x0000000000000000
0000000000000258:               0x0000000000000000
000000000000025c:               0x0000000000000000
0000000000000260:               0x0000000000000000
0000000000000264:               0x0000000000000000
0000000000000268:               0x0000000000000000
000000000000026c:               0x0000000010000000
0000000000000270:               0x0000000000000000
0000000000000274:               0x0000000000000000
0000000000000278:               0x0000000000000000
000000000000027c:               0x0000000000000000

1.使用RTL8127网卡，UEFI固定出现以上报错，导致无法加载启动内核

2.横向对比Mellanox的CX-5，CX-6以及Intel的E810网卡，都是正常启动

3.RTL8127的网卡有做更换测试，目前判断不是单张网卡的硬件问题，而是整个8127系列均会出现的问题

4.继续使用出问题的R8127测试，发现进入UEFI后再插入网卡，也就是说让UEFI先扫描完设备，再插入网卡继续加载内核即可正常识别设备并挂载驱动，而且测试功能是正常的

Question：

以上错误的具体解读与怎么修复？

4 
--- a/Silicon/NVIDIA/Drivers/PcieControllerDxe/PcieControllerDxe.c
+++ b/Silicon/NVIDIA/Drivers/PcieControllerDxe/PcieControllerDxe.c
@@ -86,6 +86,31 @@ NVIDIA_DEVICE_DISCOVERY_CONFIG  gDeviceDiscoverDriverConfig = {
   .ThreadedDeviceStart             = TRUE
 };
 
+STATIC
+BOOLEAN
+IsT264PcieC4DeviceNode (
+  IN CONST NVIDIA_DEVICE_TREE_NODE_PROTOCOL  *DeviceTreeNode
+  )
+{
+  UINT32  PcieDomain;
+
+  if ((DeviceTreeNode == NULL) || (TegraGetChipID () != T264_CHIP_ID)) {
+    return FALSE;
+  }
+
+  if (EFI_ERROR (
+        DeviceTreeGetNodePropertyValue32 (
+          DeviceTreeNode->NodeOffset,
+          "linux,pci-domain",
+          &PcieDomain
+          )))
+  {
+    return FALSE;
+  }
+
+  return PcieDomain == 4;
+}
+
 BOOLEAN  gDelayAfterPERSTRead                                                   = FALSE;
 UINT8    gDelayAfterPERST[TEGRABL_SOC_MAX_SOCKETS][TEGRABL_MAX_PCIE_PER_SOCKET] = { 0 };
 
@@ -2176,6 +2176,14 @@ DeviceDiscoveryNotify (
   PlatformType = TegraGetPlatform ();
   Status       = EFI_SUCCESS;
   PcieFound    = FALSE;
   ChipId       = TegraGetChipID ();
 
+  if (((Phase == DeviceDiscoveryDeviceTreeCompatibility) ||
+       (Phase == DeviceDiscoveryDriverBindingSupported)) &&
+      IsT264PcieC4DeviceNode (DeviceTreeNode))
+  {
+    DEBUG ((DEBUG_INFO, "%a: skip T264 PCIe C4 in UEFI\n", __FUNCTION__));
+    return EFI_UNSUPPORTED;
+  }
+
   if (ChipId == TH500_CHIP_ID) {

测试下来通过拒绝 C4 控制器发现：更早阻止 C4 Root Bridge 和 PCI 总线枚举可以在不破坏 Linux DT 状态的前提下规避 UEFI 阶段的 C4 初始化问题

5

You need to build out UEFI debug version and use that version to check log again.

Release mode binary won’t print any detail.

However, I think the quick way is just skip PCIe controller enumeration in UEFI.

There are lots of kinds of devices that might lead to unexpected bug in UEFI. Skipping enumeration is a common way to skip those issues.

6

感谢回复，确实忽略掉UEFI的枚举后功能正常了