如何解决解析固件时，是什么导致此内核出现故障？
我正在调整一个内核模块，该模块注册一堆struct software_node链接摄像头传感器到英特尔的ipu3基础架构。注册完成后，模块将触发manifest documentation。在重新探测期间，它将解析固件以建立我需要的连接。在固件解析期间，当内核尝试调用fwnode.ops
[ 639.079642] kernel tried to execute NX-protected page - exploit attempt? (uid: 0)
[ 639.079652] BUG: unable to handle page fault for address: ffffffffc11860e0
[ 639.079658] #PF: supervisor instruction fetch in kernel mode
[ 639.079665] #PF: error_code(0x0011) - permissions violation
[ 639.079671] PGD 140a0f067 P4D 140a0f067 PUD 140a11067 PMD 13e66a067 PTE 8000000027838061
[ 639.079686] Oops: 0011 [#1] SMP PTI
[ 639.079696] CPU: 1 PID: 5259 Comm: insmod Tainted: G A C OE 5.8.0-rc7-debug #83
[ 639.079703] Hardware name: LENOVO 80U1/INVALID,BIOS 2KCN25WW 10/26/2016
[ 639.079719] RIP: 0010:surface_camera_exit+0x790/0x6b0 [surface_camera]
[ 639.079728] Code: 66 72 65 71 75 65 6e 63 79 00 62 75 73 2d 74 79 70 65 00 63 6c 6f 63 6b 2d 6c 61 6e 65 73 00 64 61 74 61 2d 6c 61 6e 65 73 00 <70> 6f 72 74 30 00 65 6e 64 70 6f 69 6e 74 30 00 01 36 46 6f 75 6e
[ 639.079736] RSP: 0018:ffffa730c1fdb8b8 EFLAGS: 00010282
[ 639.079744] RAX: ffff9228f877b1d8 RBX: ffff9228f877b1d8 RCX: 0000000000000027
[ 639.079750] RDX: ffffffffc11860e0 RSI: ffff9229eec98cd0 RDI: ffff9228f877b1d8
[ 639.079756] RBP: ffffa730c1fdb8c0 R08: 0000000000000752 R09: 0000000000000004
[ 639.079763] R10: 000000000000000a R11: 0000000000000001 R12: ffff9228f877b190
[ 639.079769] R13: ffff9228f877b208 R14: ffff9228f877ba08 R15: ffff9228f877b708
[ 639.079777] FS: 00007f4c54802540(0000) GS:ffff9229eec80000(0000) knlGS:0000000000000000
[ 639.079784] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 639.079791] CR2: ffffffffc11860e0 CR3: 000000003f7f2003 CR4: 00000000003606e0
[ 639.079797] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[ 639.079803] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
[ 639.079809] Call Trace:
[ 639.079822] ? fwnode_handle_get+0x29/0x30
[ 639.079834] software_node_get_next_child+0x99/0x1f0
[ 639.079844] software_node_graph_get_next_endpoint+0xb8/0x140
[ 639.079854] fwnode_graph_get_endpoint_by_id+0x80/0x1b0
[ 639.079870] cio2_pci_probe+0x64c/0xb55 [ipu3_cio2]
[ 639.079890] local_pci_probe+0x47/0xa0
[ 639.079900] ? local_pci_probe+0x47/0xa0
[ 639.079913] pci_device_probe+0x10b/0x1b0
[ 639.079924] really_probe+0xf5/0x440
[ 639.079934] driver_probe_device+0xe8/0x150
[ 639.079943] __device_attach_driver+0x7b/0xe0
[ 639.079952] ? driver_allows_async_probing+0x60/0x60
[ 639.079963] bus_for_each_drv+0x6e/0xb0
[ 639.079972] __device_attach+0xd8/0x160
[ 639.079984] device_attach+0x10/0x20
[ 639.079991] bus_rescan_devices_helper+0x3a/0x80
[ 639.079998] device_reprobe+0x23/0x30
[ 639.080010] surface_camera_init+0x1bb/0x250 [surface_camera]
[ 639.080022] ? surface_camera_unregister_nodes+0x110/0x110 [surface_camera]
[ 639.080034] do_one_initcall+0x4a/0x200
[ 639.080047] ? _cond_resched+0x19/0x40
[ 639.080059] ? kmem_cache_alloc_trace+0x165/0x220
[ 639.080072] do_init_module+0x5f/0x21a
[ 639.080080] load_module+0x26a5/0x2d30
[ 639.080099] __do_sys_finit_module+0xfc/0x120
[ 639.080107] ? __do_sys_finit_module+0xfc/0x120
[ 639.080121] __x64_sys_finit_module+0x1a/0x20
[ 639.080128] do_syscall_64+0x48/0xc0
[ 639.080139] entry_SYSCALL_64_after_hwframe+0x44/0xa9
[ 639.080148] RIP: 0033:0x7f4c5433e959
[ 639.080158] Code: 00 f3 c3 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 40 00 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 8b 0d ff f4 2c 00 f7 d8 64 89 01 48
[ 639.080165] RSP: 002b:00007ffd0b94ad08 EFLAGS: 00000246 ORIG_RAX: 0000000000000139
[ 639.080174] RAX: ffffffffffffffda RBX: 000055699a8587c0 RCX: 00007f4c5433e959
[ 639.080180] RDX: 0000000000000000 RSI: 000055699a1decee RDI: 0000000000000003
[ 639.080187] RBP: 000055699a1decee R08: 0000000000000000 R09: 00007f4c54611000
[ 639.080192] R10: 0000000000000003 R11: 0000000000000246 R12: 0000000000000000
[ 639.080197] R13: 000055699a85b190 R14: 0000000000000000 R15: 0000000000000000
[ 639.080207] Modules linked in: surface_camera(OE+) ov5648(OE) ov2680(OE) rfcomm xt_conntrack xt_MASQUERADE nf_conntrack_netlink nfnetlink xfrm_user xfrm_algo xt_addrtype iptable_filter iptable_nat nf_nat nf_conntrack nf_defrag_ipv6 nf_defrag_ipv4 libcrc32c br_netfilter bridge stp llc ccm hid_logitech_hidpp overlay cmac bnep btusb btrtl btbcm btintel bluetooth hid_logitech_dj hid_multitouch ecdh_generic ecc binfmt_misc nls_iso8859_1 dm_crypt mousedev joydev wacom clk_tps68470 tps68470_regulator usbhid hid_generic mei_hdcp intel_rapl_msr snd_soc_skl snd_hda_codec_hdmi snd_soc_sst_ipc snd_soc_sst_dsp snd_hda_ext_core snd_soc_acpi_intel_match snd_soc_acpi snd_hda_codec_realtek snd_soc_core snd_hda_codec_generic x86_pkg_temp_thermal ledtrig_audio intel_powerclamp snd_compress coretemp ac97_bus snd_pcm_dmaengine kvm_intel snd_hda_intel snd_intel_dspcfg snd_hda_codec kvm snd_hda_core snd_hwdep snd_pcm crct10dif_pclmul crc32_pclmul ghash_clmulni_intel snd_seq_midi
[ 639.080298] snd_seq_midi_event snd_rawmidi snd_seq aesni_intel crypto_simd cryptd snd_seq_device glue_helper snd_timer rapl iwlmvm intel_cstate mac80211 libarc4 input_leds serio_raw iwlwifi snd wmi_bmof efi_pstore soundcore intel_wmi_thunderbolt cfg80211 mei_me ipu3_imgu(C) ipu3_cio2 processor_thermal_device mei intel_lpss_pci v4l2_fwnode intel_lpss videobuf2_dma_sg idma64 videobuf2_memops virt_dma videobuf2_v4l2 bmc150_accel_i2c videobuf2_common intel_rapl_common bmc150_accel_core intel_xhci_usb_role_switch intel_pch_thermal int340x_thermal_zone roles videodev industrialio_triggered_buffer kfifo_buf mc intel_soc_dts_iosf industrialio intel_vbtn sparse_keymap int3400_thermal acpi_thermal_rel acpi_pad mac_hid sch_fq_codel parport_pc ppdev lp parport ip_tables x_tables autofs4 i915 i2c_algo_bit drm_kms_helper syscopyarea sysfillrect sysimgblt fb_sys_fops cec rc_core nvme ahci nvme_core drm libahci xhci_pci xhci_pci_renesas i2c_hid hid wmi video pinctrl_sunrisepoint
[ 639.080391] pinctrl_intel backlight
[ 639.080404] CR2: ffffffffc11860e0o
[ 639.080410] ---[ end trace c651565dd757a731 ]---
[ 639.366643] RIP: 0010:surface_camera_exit+0x790/0x6b0 [surface_camera]
[ 639.366647] Code: 66 72 65 71 75 65 6e 63 79 00 62 75 73 2d 74 79 70 65 00 63 6c 6f 63 6b 2d 6c 61 6e 65 73 00 64 61 74 61 2d 6c 61 6e 65 73 00 <70> 6f 72 74 30 00 65 6e 64 70 6f 69 6e 74 30 00 01 36 46 6f 75 6e
[ 639.366649] RSP: 0018:ffffa730c1fdb8b8 EFLAGS: 00010282
[ 639.366651] RAX: ffff9228f877b1d8 RBX: ffff9228f877b1d8 RCX: 0000000000000027
[ 639.366653] RDX: ffffffffc11860e0 RSI: ffff9229eec98cd0 RDI: ffff9228f877b1d8
[ 639.366654] RBP: ffffa730c1fdb8c0 R08: 0000000000000752 R09: 0000000000000004
[ 639.366655] R10: 000000000000000a R11: 0000000000000001 R12: ffff9228f877b190
[ 639.366657] R13: ffff9228f877b208 R14: ffff9228f877ba08 R15: ffff9228f877b708
[ 639.366659] FS: 00007f4c54802540(0000) GS:ffff9229eec80000(0000) knlGS:0000000000000000
[ 639.366660] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 639.366662] CR2: ffffffffc11860e0 CR3: 000000003f7f2003 CR4: 00000000003606e0
[ 639.366663] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[ 639.366664] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
现在，我的笔记本电脑有2个传感器。注册其中任何一个模块时，该模块都能正常工作（它通过ACPI手柄检测到的传感器，但是在dev->driver_data中找不到任何数据，它只是忽略了该传感器，因此modprobe -r的传感器模块让我确认它对于一个传感器）。因此，大概我只是以某种方式搞砸了注册，但是经过几天的尝试，我看不到自己做错了什么。在调试工作中，我得出的结论是，遇到的 first 传感器已正确注册（这是有意义的，因为它们是独立工作的）。在应用了一些补丁之后，由fwnode_handle_get触发对software_node_get_next_child的调用：请参见以下链接：
static struct fwnode_handle *
software_node_get_next_child(const struct fwnode_handle *fwnode,struct fwnode_handle *child)
{
struct swnode *p = to_swnode(fwnode);
struct swnode *c = to_swnode(child);
if (!p || list_empty(&p->children) ||
(c && list_is_last(&c->entry,&p->children)))
return NULL;
if (c)
c = list_next_entry(c,entry);
else
c = list_first_entry(&p->children,struct swnode,entry);
return fwnode_handle_get(&c->fwnode);
}
据我所知，由于c = list_next_entry(c,entry)返回了无效的指针，因此发生了错误。我为此添加了一些调试打印：
pr_info("Before list_next_entry\n");
pr_info(" c->node is called %s\n",c->node->name);
pr_info(" its address is %d\n",c->node);
pr_info(" addr of c->fwnode is %d\n",&c->fwnode);
c = list_next_entry(c,entry);
pr_info("After list_next_entry\n");
pr_info(" addr of c: %d\n",c);
pr_info(" addr of c->node: %d\n",c->node);
// pr_info("addr of c->node->name: %d\n",c->node->name);
pr_info(" addr of c->fwnodw: %d\n",&c->fwnode);
[ 639.079614] Before list_next_entry
[ 639.079616] c->node is called endpoint0
[ 639.079618] its address is -1055361936
[ 639.079619] addr of c->fwnode is -126372088
[ 639.079621] After list_next_entry
[ 639.079624] addr of c: -126373488
[ 639.079625] addr of c->node: 3
[ 639.079627] addr of c->fwnodw: -126373416
“ 3”不是有效的内存地址，因此无论swnode是由什么构成的，它似乎都不是我最初注册的software_nodes之一。但是在那种情况下，我不知道为什么list_next_entry会返回它。实际上，根据我的估计，应该从未遇到过这段代码：list_is_last(&c->entry,&p->children)应该是true，因为我每个端口只注册一个孩子。 c->node->name节点返回endpoint0。
完整的模块相当大，很难最小化，但是在我仍然可以重现错误的同时，它可以做到的很小。但是请注意，如果您想实际运行代码，则需要一台带有cio2设备的设备，该设备使用ipu3驱动程序以及两个传感器（您需要用其替换为supported_devices数组的ACPI HID）来重现它。瞧瞧。
#include <linux/acpi.h>
#include <acpi/acpi_bus.h>
#include <linux/device.h>
#include <linux/i2c.h>
#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/pci.h>
#include <media/v4l2-subdev.h>
#include <linux/fwnode.h>
#define MAX_CONNECTED_DEVICES 4
#define SWNODE_SENSOR_HID 0
#define SWNODE_SENSOR_PORT 1
#define SWNODE_SENSOR_ENDPOINT 2
#define SWNODE_CIO2_PORT 3
#define SWNODE_CIO2_ENDPOINT 4
#define SWNODE_NULL_TERMINATOR 5
#define CIO2_HID "INT343E"
#define CIO2_PCI_ID 0x9d32
#define ENDPOINT_SENSOR 0
#define ENDPOINT_CIO2 1
#define NODE_HID(_HID) \
(const struct software_node) { \
_HID,\
}
#define NODE_PORT(_PORT,_HID_NODE) \
(const struct software_node) { \
_PORT,\
_HID_NODE,\
}
#define NODE_ENDPOINT(_EP,_PORT,_PROPS) \
(const struct software_node) { \
_EP,\
_PORT,\
_PROPS,\
}
#define PROPERTY_ENTRY_NULL \
(const struct property_entry) { }
#define SOFTWARE_NODE_NULL \
(const struct software_node) { }
static char* supported_devices[] = {
"OVTI5648","OVTI2680",};
const char *port_names[] = {
"port0","port1"
};
struct pci_dev *cio2;
struct software_node cio2_hid_node = { CIO2_HID,};
struct sensor {
struct software_node swnodes[7];
struct property_entry sensor_props[6];
struct property_entry cio2_props[3];
struct fwnode_handle *fwnode;
};
struct connected_devices {
int n_devices;
struct sensor sensors[MAX_CONNECTED_DEVICES];
};
struct connected_devices connected_devs = {
.n_devices = 0,};
static const struct property_entry remote_endpoints[] = {
PROPERTY_ENTRY_REF("remote-endpoint",&connected_devs.sensors[0].swnodes[SWNODE_CIO2_ENDPOINT]),/* Sensor 0,Sensor Property */
PROPERTY_ENTRY_REF("remote-endpoint",&connected_devs.sensors[0].swnodes[SWNODE_SENSOR_ENDPOINT]),CIO2 Property */
PROPERTY_ENTRY_REF("remote-endpoint",&connected_devs.sensors[1].swnodes[SWNODE_CIO2_ENDPOINT]),PROPERTY_ENTRY_REF("remote-endpoint",&connected_devs.sensors[1].swnodes[SWNODE_SENSOR_ENDPOINT]),{ }
};
static int connect_supported_devices(struct device *dev,void *data)
{
struct acpi_device *adev;
struct connected_devices *cdevs = data;
struct property_entry *sensor_props;
struct property_entry *cio2_props;
struct fwnode_handle *fwnode;
struct software_node *nodes;
const char *hid;
int i,ret;
adev = ACPI_COMPANION(dev);
if (!adev) {
return 0;
}
hid = acpi_device_hid(adev);
for (i = 0; i < ARRAY_SIZE(supported_devices); i++) {
if (!strcmp(hid,supported_devices[i])) {
if (!dev->driver_data) {
pr_info("Found supported device %s,but it has no driver; skipping\n",hid);
return 0;
}
nodes = cdevs->sensors[cdevs->n_devices].swnodes;
sensor_props = cdevs->sensors[cdevs->n_devices].sensor_props;
cio2_props = cdevs->sensors[cdevs->n_devices].cio2_props;
fwnode = cdevs->sensors[cdevs->n_devices].fwnode;
sensor_props[0] = remote_endpoints[(cdevs->n_devices * 2) + ENDPOINT_SENSOR];
sensor_props[1] = PROPERTY_ENTRY_NULL;
cio2_props[0] = remote_endpoints[(cdevs->n_devices * 2) + ENDPOINT_CIO2];
cio2_props[1] = PROPERTY_ENTRY_NULL;
/* build the software nodes */
nodes[SWNODE_SENSOR_HID] = NODE_HID(supported_devices[i]); /* Sensor HID Node */
nodes[SWNODE_SENSOR_PORT] = NODE_PORT("port0",&nodes[SWNODE_SENSOR_HID]); /* Sensor Port Node */
nodes[SWNODE_SENSOR_ENDPOINT] = NODE_ENDPOINT("endpoint0",&nodes[SWNODE_SENSOR_PORT],sensor_props); /* Sensor Endpoint Node */
nodes[SWNODE_CIO2_PORT] = NODE_PORT(port_names[i],&cio2_hid_node); /* CIO2 Port Node */
nodes[SWNODE_CIO2_ENDPOINT] = NODE_ENDPOINT("endpoint0",&nodes[SWNODE_CIO2_PORT],cio2_props); /* CIO2 Endpoint Node */
nodes[SWNODE_NULL_TERMINATOR] = SOFTWARE_NODE_NULL;
ret = software_node_register_nodes(nodes);
if (ret) {
dev_err(dev,"Failed to register the software nodes for %s\n",supported_devices[i]);
return 0;
}
fwnode = software_node_fwnode(&nodes[SWNODE_SENSOR_HID]);
if (!fwnode) {
dev_err(dev,"Failed to get fwnode from software node for %s\n",supported_devices[i]);
return 0;
}
//fwnode->secondary = ERR_PTR(-ENODEV);
dev->fwnode = fwnode;
((struct v4l2_subdev *)dev->driver_data)->fwnode = fwnode;
cdevs->n_devices++;
return 0;
}
}
return 0;
}
static int surface_camera_init(void)
{
struct fwnode_handle *fwnode;
int ret;
/* Register the CIO2 Parent node */
ret = software_node_register(&cio2_hid_node);
if (ret < 0) {
pr_err("Failed to register the CIO2 HID node\n");
return -EINVAL;
}
/* Check for supported devices and connect them*/
ret = i2c_for_each_dev(&connected_devs,connect_supported_devices);
if ((ret < 0) || (connected_devs.n_devices == 0)) {
pr_err("Failed to connect any devices\n");
goto out;
}
/* Find pci device and add swnode as primary */
cio2 = pci_get_device(PCI_VENDOR_ID_INTEL,CIO2_PCI_ID,NULL);
if (!cio2) {
ret = -EPROBE_DEFER;
goto out;
}
fwnode = software_node_fwnode(&cio2_hid_node);
if (!fwnode) {
pr_err("Error getting fwnode from cio2 software_node\n");
ret = -ENODEV;
goto out;
}
fwnode->secondary = ERR_PTR(-ENODEV);
cio2->dev.fwnode = fwnode;
pr_info("Reprobing now\n");
ret = device_reprobe(&cio2->dev);
if (ret) {
dev_warn(&cio2->dev,"Reprobing error: %d\n",ret);
goto out;
}
return 0;
out:
return ret;
}
module_init(surface_camera_init);
MODULE_DESCRIPTION("A bridge driver to connect sensors to CIO2 infrastructure.");
MODULE_LICENSE("GPL v2");
MODULE_ALIAS("acpi*:INT343E:*");
问题是；内核oops的根本原因是什么？修复它的最佳方法是什么？
这是针对5.8.0-rc7内核并应用了多个补丁的；尤其是reprobing the cio2 device中的1-3（补丁5是我尝试扩展的原始模块）。
我认为这是所有需要的信息，但是如果您想检查任何内容，只需询问。
解决方法
事实证明，这只是外围连接到我的代码；该错误实际上是我在问题中链接的补丁之一，现在已解决。
版权声明：本文内容由互联网用户自发贡献，该文观点与技术仅代表作者本人。本站仅提供信息存储空间服务，不拥有所有权，不承担相关法律责任。如发现本站有涉嫌侵权/违法违规的内容， 请发送邮件至 dio@foxmail.com 举报，一经查实，本站将立刻删除。
