Intel C62x芯片组QAT驱动安装失败：Exec format error与Skipping invalid relocation target错误排查求助

大家好，我最近在更新Intel C62x Chipset QuickAssist Technology（QAT）扩展卡的驱动时遇到了棘手的问题，加载模块时频繁报错。我记得之前的更新流程挺简单的，但这次卡在了安装环节，具体情况如下：

• 停止QAT服务：成功
• 卸载旧QAT驱动：成功
• 移除残留的QAT模块（modprobe -r）：成功
• 下载、解压、配置、编译新QAT驱动：成功
• 安装新驱动：失败

我执行的安装命令是：

make -j16 install

输出的关键日志片段如下：

[...]
make -C /lib/modules/4.19.0-5-amd64/build M=/usr/local/src/QAT.L.4.20.0-00001/quickassist/qat INSTALL_MOD_DIR=updates modules_install

make[3]: Entering directory '/usr/src/linux-headers-4.19.0-5-amd64'

INSTALL /usr/local/src/QAT.L.4.20.0-00001/quickassist/qat/drivers/crypto/qat/qat_200xx/qat_200xx.ko
INSTALL /usr/local/src/QAT.L.4.20.0-00001/quickassist/qat/drivers/crypto/qat/qat_200xxvf/qat_200xxvf.ko
INSTALL /usr/local/src/QAT.L.4.20.0-00001/quickassist/qat/drivers/crypto/qat/qat_c3xxx/qat_c3xxx.ko
INSTALL /usr/local/src/QAT.L.4.20.0-00001/quickassist/qat/drivers/crypto/qat/qat_c3xxxvf/qat_c3xxxvf.ko
INSTALL /usr/local/src/QAT.L.4.20.0-00001/quickassist/qat/drivers/crypto/qat/qat_c4xxx/qat_c4xxx.ko
INSTALL /usr/local/src/QAT.L.4.20.0-00001/quickassist/qat/drivers/crypto/qat/qat_c4xxxvf/qat_c4xxxvf.ko
INSTALL /usr/local/src/QAT.L.4.20.0-00001/quickassist/qat/drivers/crypto/qat/qat_c62x/qat_c62x.ko
INSTALL /usr/local/src/QAT.L.4.20.0-00001/quickassist/qat/drivers/crypto/qat/qat_c62xvf/qat_c62xvf.ko
INSTALL /usr/local/src/QAT.L.4.20.0-00001/quickassist/qat/drivers/crypto/qat/qat_common/intel_qat.ko
INSTALL /usr/local/src/QAT.L.4.20.0-00001/quickassist/qat/drivers/crypto/qat/qat_d15xx/qat_d15xx.ko
INSTALL /usr/local/src/QAT.L.4.20.0-00001/quickassist/qat/drivers/crypto/qat/qat_d15xxvf/qat_d15xxvf.ko
INSTALL /usr/local/src/QAT.L.4.20.0-00001/quickassist/qat/drivers/crypto/qat/qat_dh895xcc/qat_dh895xcc.ko
INSTALL /usr/local/src/QAT.L.4.20.0-00001/quickassist/qat/drivers/crypto/qat/qat_dh895xccvf/qat_dh895xccvf.ko

DEPMOD  4.19.0-5-amd64
Warning: modules_install: missing 'System.map' file. Skipping depmod.

make[3]: Leaving directory '/usr/src/linux-headers-4.19.0-5-amd64'
make[2]: Leaving directory '/usr/local/src/QAT.L.4.20.0-00001/quickassist/qat'

Creating startup and kill scripts
Copying libqat_s.so to /usr/local/lib
Copying libusdm_drv_s.so to /usr/local/lib
Copying usdm module to system drivers
Group qat already exists.
Creating udev rules
Creating module.dep file for QAT released kernel object
This will take a few moments
Created symlink /etc/systemd/system/multi-user.target.wants/qat.service → /lib/systemd/system/qat.service.

Starting QAT service
modprobe: ERROR: could not insert 'qat_c62x': Exec format error
modprobe: ERROR: could not insert 'usdm_drv': Exec format error
modprobe: ERROR: could not insert 'qat_api': Exec format error

Restarting all devices.
Can not open /dev/qat_adf_ctl
Checking status of all devices.
Can not open /dev/qat_adf_ctl

*** Error:usdm_drv module not installed ***
*** Error:qat_api module not installed ***
*** Error:qat_c62x module not installed ***

Can not open /dev/qat_adf_ctl

make[1]: Nothing to be done for 'install-data-am'.
make[1]: Leaving directory '/usr/local/src/QAT.L.4.20.0-00001'

我手动尝试加载模块也报同样的错误：

# modprobe -v qat_c62x
insmod /lib/modules/4.19.0-5-amd64/updates/drivers/crypto/qat/qat_common/intel_qat.ko
modprobe: ERROR: could not insert 'qat_c62x': Exec format error

# modprobe -v intel_qat
insmod /lib/modules/4.19.0-5-amd64/updates/drivers/crypto/qat/qat_common/intel_qat.ko
modprobe: ERROR: could not insert 'intel_qat': Exec format error

同时，系统日志里出现了这些奇怪的报错：

<date> <host> kernel: [<uptime>] module: x86/modules: Skipping invalid relocation target, existing value is nonzero for type 1, loc 000000005c09d38d, val ffffffffc09ba2fe
<date> <host> kernel: [<uptime>] module: x86/modules: Skipping invalid relocation target, existing value is nonzero for type 1, loc 00000000690eef73, val ffffffffc0b572fe
<date> <host> kernel: [<uptime>] module: x86/modules: Skipping invalid relocation target, existing value is nonzero for type 1, loc 000000007d30ad16, val ffffffffc0c862fe
<date> <host> kernel: [<uptime>] module: x86/modules: Skipping invalid relocation target, existing value is nonzero for type 1, loc 00000000671d2da7, val ffffffffc0cd82fe
<date> <host> kernel: [<uptime>] module: x86/modules: Skipping invalid relocation target, existing value is nonzero for type 1, loc 0000000072877855, val ffffffffc0d2a2fe

我完全搞不懂这些错误到底在说什么，想请教各位大佬：这些错误的原因是什么？有没有可行的解决办法？谢谢大家！

备注：内容来源于stack exchange，提问作者AGI-Chandler