Table of Contents |
---|
...
GPI provides a wide set of primitives through the API level called Service Abstraction Level (SAL). This SAL layer encapsulates Agnos' HAL corresponding to the DEVICE layer. Integrating a new platform means overloading DEVICE behavior, i.e. coding into HAL. However, some GPI primitives may be overloaded depending on design decisionsL3 needs:
gpiInitializeHSM (***): overload this primitive to modify file references
gpiEMVGetCertificate (***): overload this primitive to modify public keys look-up (not required for L2 TA)
gpiGetEMVCRL (***): overload this primitive to modify revocated certificates look-up (not required for L2 TA)
gpiFindPANfromEFL (***): overload this primitive to modify exception file parsing
Log and Profiling primitives (***): overload these primitives to adpat log and profiling to platform’s resources
Initialization
(not required for L2 TA)
Initialization
All these primitives shall be subject to unit tests before any functional runs. This set of primitives must be integrated in priority. Please refer to file headers to get most up to date primitives signatures.
...
Primitives | Module | Device Mapping | Comments | |||||
---|---|---|---|---|---|---|---|---|
User Interface | UI primitives are limited to a few set of mandatory functions (here below). EMV processing doesn’t require sophisticated displays. However, a minimum set of functions is required to provide a feedback to the cardholder and the merchants. Even if card processing components don’t use UI services (see exceptions however), these primitives are important to implement early in the integration process in order to be able to trace when debugging the system. There are two exec2ption related to UI services used by card processing:
| |||||||
gpiGetString | SPED | dev_str_get_string_by_reference | ||||||
gpiDisplayMessage | SPED | dev_lcd_put_string | Display a string at X and Y coordinates. gpiInitializeDisplay shall be called ahead for a proper lines/colums management | |||||
gpiDisplayMessageByID | SPED | dev_str_get_string_by_reference | Display a string from the string table indexed by ID | |||||
gpiSwitchLED | SPED | Swith on/off a LED. LED#2 is called duruing contactless card processing. LED#2 maybe display along with LED#3 if performance concerns. Example:
| ||||||
gpiAdvancedDisplay | SPED | |||||||
gpiAdvancedDisplayAsOutcome | SPED | |||||||
gpiSetPollingMessage | SPED | N/A | Set the message (up to three lines) to be displayed at AgnosEP’s polling time | |||||
gpiDisplayPollingMessage | SPED | dev_lcd_put_string | Display the message (up to three lines) - at AgnosEP’s polling time - that has been set from gpiSetPollingMessage | |||||
Memory | Platform/Memory primitives are used to optimize the memory allocation for card processing. Hence, no dynamic allocation is performed within the framework. There a 6 different king of RAM memory slots: DEV_TOTAL_RAM (0x01): DEV_TAGS_RAM + DEV_UNKNOWN_TAGS_RAM + DEV_UNKNOWN_TAGS_RAM DEV_TAGS_RAM (0x02): all tags supporting the card processing coming from device configuration or the card DEV_UNKNOWN_TAGS_RAM (0x03): any tags coming out from the card and not supported by AgnosDB DEV_FREE_RAM (0x04): extra RAM to support any king of processing beyond AGF if shared RAM used beyond AGF DEV_SLOT_RAM (0x05): private RAM to store data to support a proprietary processing such as torn mechanism. As of tofay, C2/MCW and C5/JCB requires these pads. DEV_DO_RAM (0x06): all tags corresponding to the CL combination in progress. This RAM space is used to load ENTRY POINT data (for a given transaction type) in order to prepare entry point confugration data specific to the transaction in progress. Recommended sizes: DEV_TAGS_RAM = 11500 bytes DEV_UNKNOWN_TAGS_RAM = 2500 bytes DEV_FREE_RAM = 3000 bytes DEV_SLOT_RAM = 17100 bytes DEV_DO_RAM = 10000 bytes | |||||||
gpiGetTotalRAM | Platform | dev_get_shared_ram_size dev_get_shared_ram | Provide a pointer and the corresponding available size | |||||
gpiGetRAMForTags | Platform | dev_get_shared_ram_size dev_get_shared_ram | Provide a pointer and the corresponding available sizev | |||||
gpiGetRAMForUnknownTags | Platform | dev_get_shared_ram_size dev_get_shared_ram | Provide a pointer and the corresponding available size | |||||
gpiGetFreeRAM | Platform | dev_get_shared_ram_size dev_get_shared_ram | Provide a pointer and the corresponding available size | |||||
gpiGetPrivateScratchPAD | Platform | dev_get_shared_ram_size dev_get_shared_ram | Provide a pointer and the corresponding available size | |||||
gpiGetDataObjectRAM | Platform | dev_get_shared_ram_size dev_get_shared_ram | Provide a pointer and the corresponding available size | |||||
gpiMemCpy | Platform | dev_mem_copy | Standard C . See <string.h> | |||||
gpiMemCmp | Platform | dev_mem_cmp | Standard C . See <string.h> | |||||
gpiMemSet | Platform | dev_mem_set | Standard C . See <string.h> | |||||
gpiStrCpy | Platform | dev_str_copy | Standard C . See <string.h> | |||||
gpiStrNCpy | Platform | dev_str_n_copy | Standard C . See <string.h> | |||||
gpiStrCmp | Platform | dev_str_cmp | Standard C . See <string.h> | |||||
gpiStrLen | Platform | dev_str_len | Standard C . See <string.h> | |||||
RTC | Timers Timers are used for two purposes:
Becareful, timers rely on dev_rtc_get_accurate_time_of_day implementation. | |||||||
gpiGetDate | Platform | dev_rtc_get_date | ||||||
gpiGetTime | Platform | dev_rtc_get_time | ||||||
gpiSetDateTime | Platform | dev_rtc_set_date dev_rtc_set_time | ||||||
gpiResetTimers | Platform | N/A | Free up and reset all the timers | |||||
gpiGetMaxTimer | Platform | N/A | Maximum number of timers is set by default to 3 in the GPI (see rtc.c) | |||||
gpiGetTimer | Platform | N/A | Provide a free timer for further processing | |||||
gpiStartTimer | Platform | dev_rtc_get_accurate_time_of_day | ||||||
gpiRestartTimer | Platform | dev_rtc_get_accurate_time_of_day | ||||||
gpiStopTimer | Platform | dev_rtc_get_accurate_time_of_day | ||||||
gpiGetTimerTime | Platform | dev_rtc_get_accurate_time_of_day | ||||||
gpiGetTimerState | Platform | N/A | ||||||
gpiFreeTimer | Platform | N/A | Reset a specific timer | |||||
gpiGetLastAllocated | Platform | N/A | Return the very last timer that was used | |||||
gpiSleep | Platform | dev_rtc_sleep | ||||||
HSM | ||||||||
gpiRSAComputation | HSM | dev_hsm_rsa_computation | See gpihsm.h. This primitive is used from inside EMV core engine for ODA | |||||
gpiShaInit | HSM | dev_hsm_sha_init | See gpihsm.h. This primitive may ne used to calculate any SHA-1 based signatures | |||||
gpiShaUpdate | HSM | dev_hsm_sha_update | See gpihsm.h. This primitive may ne used to calculate any SHA-1 based signatures | |||||
gpiShaFinal | HSM | dev_hsm_sha_final | See gpihsm.h. This primitive may ne used to calculate any SHA-1 based signatures | |||||
gpiGetChecksum | HSM | dev_hsm_get_checksum | See gpihsm.h. Provide the checksum corresponding to each L2 CT/CL kernel | |||||
gpiGetRandomNumber | HSM | dev_hsm_get_random_number | See gpihsm.h.. Generate a random are between 0 and 255 | |||||
gpiGetEMVCertificate (***) | HSM | dev_file_open dev_file_read dev_file_close | ||||||
gpiGetEMVCRL (***) | HSM | dev_file_open dev_file_read dev_file_close | ||||||
gpiFindPANFromEFL (***) | HSM | dev_file_open dev_file_read dev_file_close | ||||||
xgpiApplicationBasedDES | xGPI | |||||||
xgpiAES128CBCEncrypt | xGPI | |||||||
File Management | Files are used to support the persistent model. However, using a file system is not requirement, and a dedicated flash memory management may be implemented instead. CONFIGURATION files:
DRL files:
UPDATE CONDITIONS files:
KEY files:
REVOCATED Files:
EXCEPTION Files:
| |||||||
gpiFileOpen | Platform | dev_file_open | Open a file. Provide a handler that can be used further for read/write/close operations | |||||
gpiFileClose | Platform | dev_file_close | Close a file | |||||
gpiFileRead | Platform | dev_file_read | Read from a file | |||||
gpiFileWrite | Platform | dev_file_write | Write in a file | |||||
gpiFileDelete | Platform | dev_file_delete | Delete a file | |||||
gpiFileRename | Platform | dev_file_rename | Rename a file | |||||
gpiFileGetSize | Platform | dev_file_get_size | Provide file’s size |
...
Primitives | Module | Device Mapping | Comment |
---|---|---|---|
Log | GPI proposes a logging mechanism that suits to profiling. It is possible to hook these cases to trace the framework without having to modify internal software componentsimportant to enable _AGNOSTRACE_ at compilation time to get his feature (for qualification and TA only) | ||
gpiInitializeLogger (***) | Platform | N/A | Open logging whereever it takes place |
gpiReleaseLogger (***) | Platform | N/A | Close logging whereever it took place |
gpiLog (***) | Platform | N/A | Overload this primitive if required |
gpiLogf (***) | Platform | N/A | Overload this primitive if required |
gpiLogData (***) | Platform | N/A | Overload this primitive if required |
gpiLogSpecific :
| |||
gpiReleaseLogger (***) | Platform | N/AOverload this primitive if required | Close logging whereever it took place |
Profiling | Profiling may be used for differential timing analysis. Absolute measurement is difficult because logging (or time stamping) takes CPU time. However, this mechanism allows to measure:
However, time stampng is performed in RAM (generic implementation. see log.c) if available RAM allows it | ||
gpiGetTimeStamp (***) | Platform | N/A | Profile CAD/gpiExchangeAPDU is the only GPI service thart that is profiled. Each APDU command/reponse time is measyred measured in order to discriminate the time spent in L1 and L2Overload this primitive directly in GPI/Platform to log whereever it is possible (generic implementation logs in RAM) Generic implementation uses clock_gettime (see log.c). Replace this function with appropriate primitive available on the platform |
gpiSwithOnTimeStamp | Platform | N/A | Activate time stamp |
gpiSwithOffTimeStamp | Platform | N/A | Deactivate time stamp |
gpiDumpTimeStamp (***) | Platform | N/A | Dump time stamps whereever possible |