如何理解这个 dmesg 错误信息？

Question

我编写了这个简单的模块来处理设备并调用它的一些电源管理方法，例如.suspend 和.resume。在初始化时，模块会简单地查找特定设备并尝试调用其方法。

#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/device.h>
#include <linux/pci.h>

static int __init mfps_driver_init(void){

struct pci_dev    *dev      = NULL;
struct pci_driver *driver   = NULL;
struct device     *device   = NULL;

dev = pci_get_device(0x8086, 0x15a2, NULL);

if((dev == NULL) || (dev == 0)){

    printk(KERN_INFO "LEONZO: NOTHING FOUND SIZE %ld\n", sizeof(dev));

} else {

    driver = dev->driver;

    printk(KERN_INFO "LEONZO: I FOUND THE DEVICE OF THE SIZE %ld\n", sizeof(dev));
    printk(KERN_INFO "LEONZO: HERE IS ITS DRIVER NAME %s\n", driver->name);
    printk(KERN_INFO "LEONZO: CALLING IT SUSPEND METHOD\n");

    *device = dev->dev;

    device_lock(device);

    device_unlock(device);
}

return 0;

}

static void __exit mfps_driver_exit(void){

}


module_init(mfps_driver_init);
module_exit(mfps_driver_exit);

代码编译成功。但是当我加载模块时出现内核错误：

sudo insmod MyFirstPowerState.ko

dmesg 显示如下输出

[   59.545180] MyFirstPowerState: module license 'unspecified' taints   kernel. 
[   59.545183] Disabling lock debugging due to kernel taint
[   59.546010] LEONZO: I FOUND THE DEVICE OF THE SIZE 8
[   59.546012] LEONZO: HERE IS ITS DRIVER NAME e1000e
[   59.546013] LEONZO: CALLING IT SUSPEND METHOD
[   59.546021] BUG: unable to handle kernel NULL pointer dereference         at           (null)
[   59.546051] IP: [<ffffffffc011907e>] mfps_driver_init+0x7e/0x1000         [MyFirstPowerState]
[   59.546077] PGD 0 
[   59.546085] Oops: 0002 [#1] SMP 
[   59.546097] Modules linked in: MyFirstPowerState(POE+) xt_CHECKSUM arc4 iwlmvm mac80211 snd_hda_codec_hdmi snd_hda_codec_realtek iwlwifi snd_hda_codec_generic rtsx_pci_ms memstick cfg80211 nf_conntrack_netbios_ns nf_conntrack_broadcast ipt_MASQUERADE nf_nat_masquerade_ipv4 xt_tcpudp ip6t_REJECT nf_reject_ipv6 ipt_REJECT nf_reject_ipv4 xt_conntrack ebtable_nat ebtable_broute bridge stp llc ebtable_filter ebtables ip6table_nat nf_conntrack_ipv6 nf_defrag_ipv6 nf_nat_ipv6 ip6table_mangle ip6table_security ip6table_raw ip6table_filter ip6_tables iptable_nat nf_conntrack_ipv4 nf_defrag_ipv4 nf_nat_ipv4 nf_nat nf_conntrack iptable_mangle iptable_security iptable_raw iptable_filter ip_tables x_tables dm_crypt hp_wmi sparse_keymap intel_rapl iosf_mbi x86_pkg_temp_thermal intel_powerclamp coretemp kvm_intel kvm crct10dif_pclmul dm_multipath crc32_pclmul scsi_dh aesni_intel aes_x86_64 lrw gf128mul glue_helper ablk_helper cryptd joydev serio_raw lpc_ich uvcvideo snd_seq_midi snd_seq_midi_event snd_rawmidi snd_hda_intel snd_hda_controller snd_hda_codec videobuf2_vmalloc snd_hwdep shpchp snd_pcm videobuf2_memops videobuf2_core v4l2_common snd_seq e1000e(OE) i915_bpo ptp mei_me pps_core mei videodev media snd_seq_device intel_ips snd_timer drm_kms_helper drm btusb snd i2c_algo_bit soundcore 8250_fintek hp_accel lis3lv02d input_polldev tpm_infineon hp_wireless mac_hid parport_pc ppdev lp parport rfcomm bnep bluetooth binfmt_misc btrfs xor raid6_pq dm_mirror dm_region_hash dm_log uas usb_storage hid_generic usbhid hid rtsx_pci_sdmmc ahci psmouse libahci rtsx_pci wmi video
[   59.546577] CPU: 1 PID: 4180 Comm: insmod Tainted: P           OE   3.19.0-51-generic #58~14.04.1-Ubuntu
[   59.546613] Hardware name: Hewlett-Packard HP EliteBook 840 G2/2216, BIOS M71 Ver. 01.05 03/26/2015
[   59.546648] task: ffff880241a7b110 ti: ffff880242f68000 task.ti: ffff880242f68000
[   59.546678] RIP: 0010:[<ffffffffc011907e>]  [<ffffffffc011907e>] mfps_driver_init+0x7e/0x1000 [MyFirstPowerState]
[   59.546720] RSP: 0018:ffff880242f6bd18  EFLAGS: 00010246
[   59.546741] RAX: 0000000000000000 RBX: ffff880245b4d000 RCX: 00000000000000ae
[   59.546772] RDX: 0000000000000000 RSI: ffff880245b4d098 RDI: 0000000000000000
[   59.546807] RBP: ffff880242f6bd28 R08: 000000000000000a R09: 0000000000000000
[   59.546839] R10: 0000000000000d53 R11: ffff880242f6b9de R12: ffffffffc06a8000
[   59.546868] R13: 0000000000000000 R14: ffffffffc0119000 R15: ffff880242f6bef8
[   59.546900] FS:  00007f8787aa6740(0000) GS:ffff88024f440000(0000) knlGS:0000000000000000
[   59.546921] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[   59.546936] CR2: 0000000000000000 CR3: 0000000244393000 CR4: 00000000003407e0
[   59.546955] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[   59.546978] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
[   59.547006] Stack:
[   59.547014]  ffffffff81c1d060 ffff880204cd3280 ffff880242f6bda8 ffffffff81002144
[   59.547046]  0000000000000001 0000000000000002 ffff8801f8ddc4c0 0000000000000001
[   59.547079]  ffff880242f6bd88 ffffffff811cef19 ffffffff810f7aac 0000000000000018
[   59.547114] Call Trace:
[   59.547131]  [<ffffffff81002144>] do_one_initcall+0xd4/0x210
[   59.547162]  [<ffffffff811cef19>] ? kmem_cache_alloc_trace+0x199/0x220
[   59.547194]  [<ffffffff810f7aac>] ? load_module+0x164c/0x1cc0
[   59.547222]  [<ffffffff810f7ae5>] load_module+0x1685/0x1cc0
[   59.547247]  [<ffffffff810f3380>] ? store_uevent+0x40/0x40
[   59.547274]  [<ffffffff810f8296>] SyS_finit_module+0x86/0xb0
[   59.547298]  [<ffffffff817b788d>] system_call_fastpath+0x16/0x1b
[   59.547314] Code: c7 80 c0 4b c0 31 c0 e8 19 14 69 c1 48 c7 c7 a8 c0  4b c0 31 c0 e8 0b 14 69 c1 31 c0 48 8d b3 98 00 00 00 b9 ae 00 00 00 48 89 c7 <f3> a5 bf 60 00 00 00 e8 26 c7 69 c1 bf 60 00 00 00 e8 ac c5 69 
[   59.547393] RIP  [<ffffffffc011907e>] mfps_driver_init+0x7e/0x1000 [MyFirstPowerState]
[   59.547416]  RSP <ffff880242f6bd18>
[   59.547425] CR2: 0000000000000000
[   59.554577] ---[ end trace 42e3b1c73677cdfa ]---

我还注意到，因此无法删除该模块：

sudo rmmod MyFirstPowerState.ko 
rmmod: ERROR: Module MyFirstPowerState is in use

知道这段代码的含义以及如何纠正错误吗？

Answer 1

我将尝试解释下面是 dmesg 的巨大文本墙。请注意，左侧括号中的值是我忘记它们的确切关系的时间，但对你来说它们并不重要。

[59.545180] MyFirstPowerState：模块许可证“未指定”污染内核。 [59.545183]由于内核污染而禁用锁定调试

这是因为您没有声明模块许可证。通常你会看到人们在他们的代码中与 module_init 放在同一部分中。

MODULE_LICENSE("GPL");

[ 59.546010] LEONZO：我找到了 8 号设备 [59.546012] LEONZO：这是它的驱动程序名称 e1000e [59.546013] LEONZO：称之为暂停方法

这些是您的 printk 消息，这里没什么特别的。

[59.546021]BUG：无法处理内核NULL指针取消引用（null）

这就是导致崩溃的真正原因所在。内核试图取消引用导致段错误的 NULL 指针。有关具体含义的更多详细信息，请参阅here。正如 Ian 之前在 cmets 中指出的那样，您崩溃的原因似乎是您将 *device=dev->dev 而不是 device=dev->dev. 在代码中您尝试将值设备点分配给 dev->dev 但是因为当前 device=NULL您试图取消引用导致崩溃的 NULL。

[59.546051] IP：[] mfps_driver_init+0x7e/0x1000 [MyFirstPowerState] [59.546648]任务：ffff880241a7b110 ti：ffff880242f68000 task.ti：ffff880242f68000

上述错误中包含的大量错误目前对您没有太大价值，对于已经部署了某些东西并且某些特定用户有问题的人来说更是如此。它列出了诸如已安装的硬件、导致崩溃的模块以及调用所有在您的案例中众所周知的东西的模块。

[ 59.546678] RIP: 0010:[] [] mfps_driver_init+0x7e/0x1000 [MyFirstPowerState][ 59.547079] ffff880242f6bd88 ffffffff811cef19 ffffffff810f7aac 0000000000000018

本节中的所有内容都是组装信息，如果您没有组装经验，这些信息对您来说毫无意义，尽管我建议您了解一些基础知识，这在这些情况下会有所帮助。上半部分是寄存器及其当前值，下半部分是当前堆栈帧。

> [   59.547114] Call Trace:
[   59.547131]  [<ffffffff81002144>] do_one_initcall+0xd4/0x210
[   59.547162]  [<ffffffff811cef19>] ? kmem_cache_alloc_trace+0x199/0x220
[   59.547194]  [<ffffffff810f7aac>] ? load_module+0x164c/0x1cc0

调用跟踪中的所有内容都非常有用，尤其是当模块变得很长且难以通过中断等内容进行调试时。基本上，它列出了系统为导致此崩溃而进行的每个函数调用（或其他）。在您的情况下，由于您从加载模块直接进入崩溃，因此跟踪实际上只有您的 load_module 以及一些包装器和一些深层系统调用。但是，如果说您的加载模块调用了另一个函数并导致崩溃，您可以在此处看到此调用路径。

最后一点似乎是更多的寄存器。

希望这解释了当您导致内核问题时从 dmesg 获得的文本墙（不确定这是否是恐慌，请纠正我）。如果还有什么不清楚的地方，我会尽力解释，尽管我绝不是这方面的专家。