PCIe网口yt6801调试总结

精选原创

707307207_cto 2024-01-26 15:01:44 ©著作权

©著作权归作者所有：来自51CTO博客作者707307207_cto的原创作品，请联系作者获取转载授权，否则将追究法律责任

第一步：

确定 pinmap 配置是否正确

diff --git a/target/ums9620_2h10/pinmap.c b/target/ums9620_2h10/pinmap.c
index 36624548..7fe0ed8c 100755
--- a/target/ums9620_2h10/pinmap.c
+++ b/target/ums9620_2h10/pinmap.c
@@ -433,11 +433,11 @@ static pinmap_t pinmap[]={
{REG_MISC_PIN_SD1_D0, BITS_PIN_DS(3)|BIT_PIN_NULL|BIT_PIN_WPU|BIT_PIN_SLP_AP|BIT_PIN_SLP_WPU|BIT_PIN_SLP_OE},//WF_SD1_D0
{REG_PIN_SD1_D1, BITS_PIN_AF(3)},
{REG_MISC_PIN_SD1_D1, BITS_PIN_DS(3)|BIT_PIN_NULL|BIT_PIN_WPU|BIT_PIN_SLP_AP|BIT_PIN_SLP_WPU|BIT_PIN_SLP_OE},//WF_SD1_D1
-{REG_PIN_SD1_CLK, BITS_PIN_AF(3)},
+{REG_PIN_SD1_CLK, BITS_PIN_AF(1)}, //PCIE3_RST
{REG_MISC_PIN_SD1_CLK, BITS_PIN_DS(4)|BIT_PIN_NULL|BIT_PIN_WPD|BIT_PIN_SLP_AP|BIT_PIN_SLP_WPD|BIT_PIN_SLP_OE},//WF_SD1_CLK
-{REG_PIN_SD1_D2, BITS_PIN_AF(3)},
+{REG_PIN_SD1_D2, BITS_PIN_AF(1)}, //PCIE3_CLK
{REG_MISC_PIN_SD1_D2, BITS_PIN_DS(3)|BIT_PIN_NULL|BIT_PIN_WPU|BIT_PIN_SLP_AP|BIT_PIN_SLP_WPU|BIT_PIN_SLP_IE},//WF_SD1_D2
-{REG_PIN_SD1_D3, BITS_PIN_AF(3)},
+{REG_PIN_SD1_D3, BITS_PIN_AF(1)}, //PCIE3_WAKEUP
{REG_MISC_PIN_SD1_D3, BITS_PIN_DS(3)|BIT_PIN_NULL|BIT_PIN_WPU|BIT_PIN_SLP_AP|BIT_PIN_SLP_WPU|BIT_PIN_SLP_OE},//WF_SD1_D3
{REG_PIN_U5TXD, BITS_PIN_AF(3)},
{REG_MISC_PIN_U5TXD, BITS_PIN_DS(1)|BIT_PIN_NULL|BIT_PIN_WPU|BIT_PIN_SLP_AP|BIT_PIN_SLP_WPU|BIT_PIN_SLP_OE},//I2C5_SCL

第二部：确定EP上电时序是否正确

PCIe网口yt6801调试总结_网口

确保AVDD，正常上电

从log看 clkreq# 都没有被拉低，正常情况下ep 启动后会拉低这个pin，针对这个问题，需要check 下 clkrq#的pinmap 配置是否正确，在平台PCIe 初始化时, EP(PCIe 外设)是否已经完成上电时序，处于正常工作状态？请先确认PCIe 相关pin的pinmap 配置是否正确； PCIe 外设的初始化阶段在哪里，最好能明确到哪行log， RC在 kernel [ 4.247078] 进行的初始化；如果 PCIe 外设在RC 前完成初始化，需要删除 dts里面 pcie1 节点的 “ep-poweron-late” 属性，如不是需要另外调用扫卡接口(有需要再提供）。

第三步：确保pcie正常工作，能正常枚举出 pci dev

前面提到的问题没有排查清楚 PCIe 是没有正常工作的， PCIe没有正常工作就没有枚举出 pci dev，外设driver 也就match 不到device .

正常枚举出 pci dev 的log 如下：

看log 已经枚举出 pci dev了，外设 driver 应该可以match了

[ 4.447631] [08-16 10:07:38.447] pci 0000:11:00.0: [1f0a:6801] type 00 class 0x020000 (2023-08-18T08:20:38Z)

第4步：确保 BAR 有地址

从log 分析，外设的 bar0 和 bar4 都是分配了地址的，请试下 void __iomem *pci_ioremap_bar(struct pci_dev *pdev, int bar) 接口能否解析出bar 对应的虚拟地址，谢谢！

[ 4.390245] [08-16 09:39:15.390] pci 0000:11:00.0: BAR 0: assigned [mem 0x880000000-0x880003fff 64bit]

[ 4.390342] [08-16 09:39:15.390] pci 0000:11:00.0: BAR 4: assigned [io 0x1000-0x10ff] (2023-08-21T07:04:33Z)

把 pci_dev 结构体的 resource 结构体所有成员变量打印出来看看

[ 281.505181]c7 [ T3144] fxgmac_probe callin

[ 281.509241]c7 [ T3144] fuxi-gmac 0000:10:00.0: enabling device (0000 -> 0003)

[ 281.516330]c7 [ T3144] resource[0] start 0x0 end 0x0 name 0000:10:00.0 flags 0 desc 0

[ 281.524104]c7 [ T3144] fxgmac_probe pci_resource_len(pcidev, 0) == 0

[ 281.530428]c7 [ T3144] resource[1] start 0x0 end 0x0 name 0000:10:00.0 flags 0 desc 0

[ 281.538212]c7 [ T3144] fxgmac_probe pci_resource_len(pcidev, 1) == 0

[ 281.544533]c7 [ T3144] resource[2] start 0x0 end 0x0 name (null) flags 0 desc 0

[ 281.551799]c7 [ T3144] fxgmac_probe pci_resource_len(pcidev, 2) == 0

[ 281.558121]c7 [ T3144] resource[3] start 0x0 end 0x0 name (null) flags 0 desc 0

[ 281.565388]c7 [ T3144] fxgmac_probe pci_resource_len(pcidev, 3) == 0

[ 281.571706]c7 [ T3144] resource[4] start 0x0 end 0x0 name (null) flags 0 desc 0

[ 281.578972]c7 [ T3144] fxgmac_probe pci_resource_len(pcidev, 4) == 0

[ 281.585293]c7 [ T3144] resource[5] start 0x0 end 0x0 name (null) flags 0 desc 0

[ 281.592568]c7 [ T3144] fxgmac_probe pci_resource_len(pcidev, 5) == 0

[ 281.598883]c7 [ T3144] resource[6] start 0x80100000 end 0x8010ffff name 0000:10:00.0 flags 287232 desc 0

第5步：确定pci dev 的vendor 和device 值

从打印的log 发现一些问题，外设driver 的match table 有问题，请问你们为何这么修改？

static const struct pci_device_id fxgmac_pci_tbl[] = {

/*{ PCI_DEVICE(PCI_VENDOR_ID_SYNOPSYS, PCI_DEVICE_FUXI_GMAC) },*/

{ PCI_DEVICE(0x16c3, 0xabcd) },

{ PCI_DEVICE(0x16c3, 0xaabc) },

{ PCI_DEVICE(0x1f0a, 0x6801) },

{ 0 }

};

请把 { PCI_DEVICE(0x16c3, 0xabcd) } 删除，这个是 rc 对应的pci dev，不是外设的pci dev(pci 0000:11:00.0: [1f0a:6801] type 00 class 0x020000)，

第6步：

确保 base mem 能正常获取到

用原来的API pcim_iomap_regions 读到base mem后，后面关于 base mem的操作，全部都失败了，所以，怀疑base mem的获取这里还是有问题的，并没有获取正确的base mem，因此后面关于 base mem的读写操作全部都失败了，请协助排查下，谢谢

第7步：

经硬件排查有一路电强制上啦后，发现yt6801 初始化成功了，但是后面open fail 使用新接口pci_ioremap_bar，可以获取到base mem reg可以正常读写成功

第8步：

/* Initialize DMA related features */
fxgmac_config_dma_bus(pdata);
printk(KERN_ALERT "%s fxgmac_config_dma_bus end\n", __func__);
fxgmac_config_osp_mode(pdata);
fxgmac_config_pblx8(pdata);
fxgmac_config_tx_pbl_val(pdata);
fxgmac_config_rx_pbl_val(pdata);
fxgmac_config_rx_coalesce(pdata);
fxgmac_config_tx_coalesce(pdata);
fxgmac_config_rx_buffer_size(pdata);
printk(KERN_ALERT "%s fxgmac_config_rx_buffer_size end\n", __func__);
fxgmac_config_tso_mode(pdata);
fxgmac_config_sph_mode(pdata);
printk(KERN_ALERT "%s fxgmac_config_sph_mode end\n", __func__);
fxgmac_config_rss(pdata);
printk(KERN_ALERT "%s fxgmac_config_rss end\n", __func__);

[ 88.866439]c3 [ T615] sipa_dummy: dev eth0 evt 14
[ 88.872320]c3 [ T615] fxgmac_open callin
[ 88.876422]c3 [ T615] alloc_channels,channel_head=000000002aecd25d,size=ch_cnt 1*704
[ 88.884236]c3 [ T615] alloc_channels,tx_ring=00000000cc177d9f,size=cnt 1*192
[ 88.891661]c6 [ T615] alloc_channels,rx_ring=00000000a92a000f,size=cnt 1*192
[ 88.898771]c6 [ T615] fxgmac_alloc_channels, for MSIx, channel 0 dma_irq=166
[ 88.906391]c6 [ T615] alloc_channels callout ok
[ 88.911732]c4 [ T615] fxgmac_alloc_ring..ch=0,tx_desc_cnt=256,rx_desc_cnt=1024
[ 88.918961]c4 [ T615] alloc_rings callout ok
[ 88.923341]c7 [ T615] fxgmac start callin here.
[ 88.928080]c7 [ T615] fxgmac_write_ephy_reg id 16 ok, ctrl=0x08100204, data=0x0000006a
[ 88.936203]c7 [ T615] fxgmac_write_ephy_reg id 30 ok, ctrl=0x081e0204, data=0x00000052
[ 88.944196]c7 [ T615] fxgmac_write_ephy_reg id 31 ok, ctrl=0x081f0204, data=0x0000231d
[ 88.952695]c7 [ T615] fxgmac_write_ephy_reg id 30 ok, ctrl=0x081e0204, data=0x00000051
[ 88.960624]c7 [ T615] fxgmac_write_ephy_reg id 31 ok, ctrl=0x081f0204, data=0x000004a9
[ 88.968579]c7 [ T615] fxgmac_write_ephy_reg id 30 ok, ctrl=0x081e0204, data=0x00000057
[ 88.976552]c7 [ T615] fxgmac_write_ephy_reg id 31 ok, ctrl=0x081f0204, data=0x0000274c
[ 88.984500]c7 [ T615] fxgmac_write_ephy_reg id 30 ok, ctrl=0x081e0204, data=0x0000a00c
[ 88.992466]c7 [ T615] fxgmac_write_ephy_reg id 31 ok, ctrl=0x081f0204, data=0x00002600
[ 89.000429]c7 [ T615] fxgmac_write_ephy_reg id 30 ok, ctrl=0x081e0204, data=0x0000a00d
[ 89.008391]c7 [ T615] fxgmac_write_ephy_reg id 31 ok, ctrl=0x081f0204, data=0x00001800
[ 89.016360]c7 [ T615] fxgmac_write_ephy_reg id 30 ok, ctrl=0x081e0204, data=0x0000a00e
[ 89.024313]c7 [ T615] fxgmac_write_ephy_reg id 31 ok, ctrl=0x081f0204, data=0x00000000
[ 89.032243]c7 [ T615] fxgmac_dismiss_all_int callin
[ 89.037223]c7 [ T615] fxgmac hw init call in
[ 89.041508]c7 [ T615] fxgmac_flush_tx_queues, reg=0x00000000b4f56789, val=0x001f0009
[ 89.049358]c7 [ T615] fxgmac_flush_tx_queues wait... reg=0x00000000b4f56789, val=0x00000000
[ 89.057695]c7 [ T615] fxgmac_hw_init Flush Tx queues end
[ 89.063060]c7 [ T615] fxgmac_hw_init fxgmac_config_dma_bus end
[ 89.068961]c7 [ T615] fxgmac_hw_init fxgmac_config_rx_buffer_size end
[ 89.075440]c7 [ T615] fxgmac_hw_init fxgmac_config_sph_mode end
[ 89.081425]c7 [ T615] <panic> p->addr = 0x0000000088bb0afd, pc = 0xffffffc0102a4338, pstate = 0x20400005
[ 89.090908]c7 [ T615] sprd_serror_debug: panic hook handler
[ 89.090982]c7 [ T615] Kernel panic - not syncing: CFI failure (target: 0xffffffc01ada106c)
[ 89.104844]c7 [ T615] CPU: 7 PID: 615 Comm: binder:615_4 Tainted: G S WC O 5.4.210-android12-9-668746-g152af218b824-dirty #30
[ 89.117045]c7 [ T615] Hardware name: Spreadtrum UMS9620 SoC (DT)
[ 89.123102]c7 [ T615] Call trace:
[ 89.126481]c7 [ T615] dump_backtrace.cfi_jt+0x0/0x4
[ 89.131499]c7 [ T615] show_stack+0x24/0x34
[ 89.135740]c7 [ T615] dump_stack+0xd8/0x158
[ 89.140063]c7 [ T615] panic+0x19c/0x42c
[ 89.144042]c7 [ T615] __ubsan_handle_cfi_check_fail_abort+0x0/0x20
[ 89.150361]c7 [ T615] __cfi_check_fail+0x0/0x20
[ 89.155082]c7 [ T615] fxgmac_hw_init+0x1384/0x13bc [yt6801]
[ 89.160751]c7 [ T615] fxgmac_start+0x108/0x508 [yt6801]
[ 89.166116]c7 [ T615] fxgmac_open+0x100/0x1e0 [yt6801]
[ 89.171394]c7 [ T615] __dev_open+0x110/0x268
[ 89.175806]c7 [ T615] __dev_change_flags+0xe0/0x238
[ 89.180823]c7 [ T615] dev_change_flags+0x40/0x84
[ 89.185583]c7 [ T615] devinet_ioctl+0x410/0x5d0
[ 89.190256]c7 [ T615] inet_ioctl+0xb8/0x1dc
[ 89.194583]c7 [ T615] sock_do_ioctl+0x68/0x1b8
[ 89.199168]c7 [ T615] sock_ioctl+0x41c/0x6b4
[ 89.203585]c7 [ T615] do_vfs_ioctl+0x3ec/0x70c
[ 89.208169]c7 [ T615] __arm64_sys_ioctl+0x7c/0xa8
[ 89.213016]c7 [ T615] el0_svc_common+0xcc/0x1bc
[ 89.217687]c7 [ T615] el0_svc_handler+0x2c/0x3c
[ 89.222363]c7 [ T615] el0_svc+0x8/0x100
[ 89.226350]c7 [ T615] SMP: stopping secondary CPUs
[ 89.231200]c1 [ C1] CPU1: stopping...
[ 89.235224]c0 [ C0] CPU0: stopping...
[ 89.239271]c2 [ C2] CPU2: stopping...
[ 89.243297]c3 [ C3] CPU3: stopping...
[ 89.247324]c6 [ C6] CPU6: stopping...
[ 89.251269]c5 [ C5] CPU5: stopping...
[ 89.255215]c4 [ C4] CPU4: stopping...
[ 89.259433]c7 [ T615] ddr_cur_freq: 0
[ 89.271212]c7 [ T615] unisoc-mailbox tx_fifo full, drop msg, dst = 3, rd =0, wt = 64
[ 89.278940]c7 [ T615] unisoc-mailbox smsg_senddie mailbox send die smsg
[ 89.285602]c7 [ T615] unisoc-mailbox smsg_senddie mailbox send die smsg
[ 89.292261]c7 [ T615] sprd-sysdump: (sysdump_panic_event) ------ in (0)
[ 90.299078]c7 [ T615] sprd-sysdump:
[ 90.302645]c7 [ T615] sprd-sysdump: *****************************************************
[ 90.310863]c7 [ T615] sprd-sysdump: * *
[ 90.319084]c7 [ T615] sprd-sysdump: * Sysdump enter, preparing debug info to dump ... *
[ 90.327306]c7 [ T615] sprd-sysdump: * *
[ 90.335525]c7 [ T615] sprd-sysdump: *****************************************************
[ 90.343747]c7 [ T615] sprd-sysdump:
[ 90.347386]c7 [ T615] sprd-sysdump: reason: CFI failure (target: 0xffffffc01ada106c), sprd_sysdump_info->crash_key: 0
[ 90.358341]c7 [ T615] sprd-sysdump: KTXT VERIFY...
[ 90.429589]c7 [ T615] sprd-sysdump: KTXT [0xffffffc010081000--0xffffffc011130000]
[ 90.437042]c7 [ T615] sprd-sysdump: SHA1:
[ 90.441111]c7 [ T615] sprd-sysdump: 3a481cfa a01f9243 32d00d64 1c8daf9b 667af798
[ 90.448552]c7 [ T615] sprd-sysdump:
[ 90.452186]c7 [ T615] sprd-sysdump: *****************************************************
[ 90.460408]c7 [ T615] sprd-sysdump: * *
[ 90.468629]c7 [ T615] sprd-sysdump: * Try to reboot system ... *
[ 90.476849]c7 [ T615] sprd-sysdump: * *
[ 90.485072]c7 [ T615] sprd-sysdump: *****************************************************
[ 90.493292]c7 [ T615] sprd-sysdump:
[ 91.497097]c7 [ T615] Kernel Offset: disabled
[ 91.501444]c7 [ T615] CPU features: 0x00030006,2a00a238
[ 91.506719]c7 [ T615] Memory Limit: none
[ 91.510703]c7 [ T615] trusty-log panic notifier - trusty version Unisoc TEE v2.1.1, qogirn6pro, Built: 00:34:16 Feb 28 2023, 277ac0c6 (builder@cishhud0028)
[ 91.535071]c7 [ T615] Rebooting in 5 seconds..

定位到 fxgmac_config_rss 这里发生了dump 请协助分析下

第9步：dump 解决

== Weihui.Zhang@unisoc.com On 2023-08-24T08:02:14Z (json) ==

[ 89.090982]c7 [ T615] Kernel panic - not syncing: CFI failure (target: 0xffffffc01ada106c)

[ 89.104844]c7 [ T615] CPU: 7 PID: 615 Comm: binder:615_4 Tainted: G S WC O 5.4.210-android12-9-668746-g152af218b824-dirty #30

[ 89.117045]c7 [ T615] Hardware name: Spreadtrum UMS9620 SoC (DT)

[ 89.123102]c7 [ T615] Call trace:

[ 89.126481]c7 [ T615] dump_backtrace.cfi_jt+0x0/0x4

[ 89.131499]c7 [ T615] show_stack+0x24/0x34

[ 89.135740]c7 [ T615] dump_stack+0xd8/0x158

[ 89.140063]c7 [ T615] panic+0x19c/0x42c

[ 89.144042]c7 [ T615] __ubsan_handle_cfi_check_fail_abort+0x0/0x20

[ 89.150361]c7 [ T615] __cfi_check_fail+0x0/0x20

[ 89.155082]c7 [ T615] fxgmac_hw_init+0x1384/0x13bc [yt6801]

[ 89.160751]c7 [ T615] fxgmac_start+0x108/0x508 [yt6801]

[ 89.166116]c7 [ T615] fxgmac_open+0x100/0x1e0 [yt6801]

[ 89.171394]c7 [ T615] __dev_open+0x110/0x268

[ 89.175806]c7 [ T615] __dev_change_flags+0xe0/0x238

[ 89.180823]c7 [ T615] dev_change_flags+0x40/0x84

[ 89.185583]c7 [ T615] devinet_ioctl+0x410/0x5d0

[ 89.190256]c7 [ T615] inet_ioctl+0xb8/0x1dc

[ 89.194583]c7 [ T615] sock_do_ioctl+0x68/0x1b8

[ 89.199168]c7 [ T615] sock_ioctl+0x41c/0x6b4

[ 89.203585]c7 [ T615] do_vfs_ioctl+0x3ec/0x70c

[ 89.208169]c7 [ T615] __arm64_sys_ioctl+0x7c/0xa8

[ 89.213016]c7 [ T615] el0_svc_common+0xcc/0x1bc

[ 89.217687]c7 [ T615] el0_svc_handler+0x2c/0x3c

[ 89.222363]c7 [ T615] el0_svc+0x8/0x100

这个CFI check失败，往往是函数指针的定义与初始化不一致引起的。比如返回值传参等不一致。void * 变成int * 等各种不规范。第三方问题，请第三方主导解决

经排查，是数组越界了，但是android11，上可以跑起来，奇怪，解决后，现在可以看到eth0
eth0 Link encap:Ethernet HWaddr 00:55:7b:b5:7d:f7 Driver fuxi-gmac
UP BROADCAST MULTICAST MTU:1500 Metric:1
RX packets:0 errors:0 dropped:0 overruns:0 frame:0
TX packets:0 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:0 TX bytes:0