[dpdk-dev] [PATCH v2] VFIO: Avoid to enable vfio while the module not loaded

[Qiu, Michael]

On 12/5/2014 6:00 PM, Burakov, Anatoly wrote:
> Hi Michael,
>
> Few nitpicks :-) (wording of the log message I guess is up to Thomas, I won't comment on that)
>
>>  lib/librte_eal/common/eal_private.h        | 36
>> ++++++++++++++++++++++++++++++
>>  lib/librte_eal/linuxapp/eal/eal_pci_vfio.c | 24 +++++++++++++++++---
>>  2 files changed, 57 insertions(+), 3 deletions(-)
>>
>> diff --git a/lib/librte_eal/common/eal_private.h
>> b/lib/librte_eal/common/eal_private.h
>> index 232fcec..e741bdb 100644
>> --- a/lib/librte_eal/common/eal_private.h
>> +++ b/lib/librte_eal/common/eal_private.h
>> @@ -35,6 +35,7 @@
>>  #define _EAL_PRIVATE_H_
>>
>>  #include <stdio.h>
>> +#include <string.h>
>>
>>  /**
>>   * Initialize the memzone subsystem (private to eal).
>> @@ -203,4 +204,39 @@ int rte_eal_alarm_init(void);
>>   */
>>  int rte_eal_dev_init(void);
>>
>> +/**
>> + * Function is to check if the kernel module(like, vfio,
>> +vfio_iommu_type1,
>> + * etc.) loaded.
>> + *
>> + * @param module_name
>> + *	The module's name which need to be checked
>> + *
>> + * @return
>> + * 	-1 means error happens(NULL pointer or open failure)
>> + * 	 0 means the module not loaded
>> + * 	 1 means the module loaded
>> + */
>> +static inline int
>> +rte_eal_check_module(const char *module_name) {
>> +	char mod_name[30]; /* Any module names can be longer than 30
>> bytes? */
>> +	int ret = 0;
>> +
>> +	if (NULL == module_name)
>> +		return -1;
>> +	FILE * fd = fopen("/proc/modules", "r");
>> +	if (fd == NULL)
>> +		return -1;
> Can we add RTE_LOG statement here, with an strerror(errno) like in other places? Fopen failed, we should at least know why :-)
>
>> +	while(!feof(fd)) {
>> +		fscanf(fd, "%s %*[^\n]", mod_name);
>> +		if(!strcmp(mod_name, module_name)) {
> Probably should use strncmp instead of strcmp.

I don't think so, if we check module "vfio", but if given  module name
is "vfio_xx", it will also correct if use strncmp.

Thanks,
Michael
>
>> +			ret = 1;
>> +			break;
>> +		}
>> +	}
>> +	fclose(fd);
>> +
>> +	return ret;
>> +}
>> +
>>  #endif /* _EAL_PRIVATE_H_ */
>> diff --git a/lib/librte_eal/linuxapp/eal/eal_pci_vfio.c
>> b/lib/librte_eal/linuxapp/eal/eal_pci_vfio.c
>> index c1246e8..52ab2d0 100644
>> --- a/lib/librte_eal/linuxapp/eal/eal_pci_vfio.c
>> +++ b/lib/librte_eal/linuxapp/eal/eal_pci_vfio.c
>> @@ -44,6 +44,7 @@
>>  #include <rte_tailq.h>
>>  #include <rte_eal_memconfig.h>
>>  #include <rte_malloc.h>
>> +#include <eal_private.h>
>>
>>  #include "eal_filesystem.h"
>>  #include "eal_pci_init.h"
>> @@ -339,10 +340,13 @@ pci_vfio_get_container_fd(void)
>>  		ret = ioctl(vfio_container_fd, VFIO_CHECK_EXTENSION,
>> VFIO_TYPE1_IOMMU);
>>  		if (ret != 1) {
>>  			if (ret < 0)
>> -				RTE_LOG(ERR, EAL, "  could not get IOMMU
>> type, "
>> -						"error %i (%s)\n", errno,
>> strerror(errno));
>> +				RTE_LOG(ERR, EAL, "  could not get IOMMU
>> type,"
>> +					" error %i (%s)\n", errno,
>> +					strerror(errno));
>>  			else
>> -				RTE_LOG(ERR, EAL, "  unsupported IOMMU
>> type!\n");
>> +				RTE_LOG(ERR, EAL, "  unsupported IOMMU
>> type! "
>> +					"expect: VFIO_TYPE1_IOMMU, "
>> +					"actual: %d\n", ret);
> I'm not even sure we need this "expected" bit at all. We don't get back the IOMMU type VFIO currently supports; rather, this code checks if VFIO's support for VFIO_TYPE1_IOMMU is enabled or not. So I would change the error message to something more descriptive, such as " required IOMMU type support not present in VFIO!\n", and get rid of the "expected".
>
>>  			close(vfio_container_fd);
>>  			return -1;
>>  		}
>> @@ -788,6 +792,20 @@ pci_vfio_enable(void)
>>  		vfio_cfg.vfio_groups[i].fd = -1;
>>  		vfio_cfg.vfio_groups[i].group_no = -1;
>>  	}
>> +
>> +	/* return error directly */
>> +	if (rte_eal_check_module("vfio") == -1 ||
>> +	    rte_eal_check_module("vfio_iommu_type1") == -1)
>> +		return -1;
>> +
>> +	/* return 0 if not all VFIO modules loaded */
>> +	if (rte_eal_check_module("vfio") == 0 ||
>> +	    rte_eal_check_module("vfio_iommu_type1") == 0) {
>> +		RTE_LOG(INFO, EAL, "VFIO modules not all loaded,"
>> +			" skip VFIO support ...\n");
>> +		return 0;
>> +	}
> Can we perhaps make one call per module instead of two? i.e. something like:
>
> int vfio_ret, vfio_ type1_ret;
> vfio_ret = rte_eal_check_module("vfio");
> vfio_type1_ret = rte_eal_check_module("vfio_iommu_type1");
>
> if (vfio_ret == -1 || vfio_type1_ret == -1)
>     return -1;
> else if (vfio_ret == 0 || vfio_type1_ret == 0) {
>      ....
>     return 0;
> }
>
>> +
>>  	vfio_cfg.vfio_container_fd = pci_vfio_get_container_fd();
>>
>>  	/* check if we have VFIO driver enabled */
>> --
>> 1.9.3
>