Changeset 284 for branch/giet_vm_ioc_drivers/giet_drivers/bdv_driver.c

Timestamp:

Jan 31, 2014, 2:37:38 PM (11 years ago)

Author:

cfuguet

Message:

Modification of comments format on SPI-SDCARD driver to respect
GIET-VM format

File:

• branch/giet_vm_ioc_drivers/giet_drivers/bdv_driver.c (modified) (2 diffs)

branch/giet_vm_ioc_drivers/giet_drivers/bdv_driver.c

@@ -1 +1 @@
 ///////////////////////////////////////////////////////////////////////////////////
-// File     : ioc_driver.c
-// Date     : 23/05/2013
-// Author   : alain greiner
+// File      : bdv_driver.c
+// Date      : 23/05/2013
+// Author    : alain greiner
+// Maintainer: cesar fuguet
 // Copyright (c) UPMC-LIP6
 ///////////////////////////////////////////////////////////////////////////////////
-// The ioc_driver.c and ioc_driver.h files are part ot the GIET-VM kernel.
+// The bdv_driver.c and bdv_driver.h files are part ot the GIET-VM kernel.
 // This driver supports the SocLib vci_block_device component, that is
 // a single channel, block oriented, external storage contrÃŽler.
@@ -14 +15 @@
 // that is always blocking, but can be called in 4 modes:
 //
-// - In BOOT_PA mode, the _bdv_access() function use the buffer virtual address 
-//   as a physical address (as the page tables are not build) and use a polling
-//   policy on the IOC_STATUS register to detect transfer completion, as
-//   hardware interrupts are not activated. This mode is used by the
-//   boot code to load the map.bin file into memory.
-//
-// - In BOOT_VA mode, the _bdv_access() function makes a V2P translation to
-//   compute the buffer physical address, and use a polling policy on IOC_STATUS
-//   register to detect transfer completion. This mode is used by the boot code
-//   to load the various .elf files into memory.
-//
-// - In KERNEL mode, the _bdv_access() function makes a V2P translation to
-//   compute the buffer physical address, and use a descheduling strategy:
-//   The ISR executed when transfer completes should restart the calling task. 
-//   There is no checking of user access right to the memory buffer.
-//   This mode must be used to access IOC, for an "open" system call.
-//
-// - In USER mode, the _bdv_access() function makes a V2P translation to
-//   compute the buffer physical address, and use a descheduling strategy:
+// - In BOOT_PA mode, the _bdv_access() function uses a polling policy on the
+//   IOC_STATUS register to detect transfer completion, as hardware interrupts
+//   are not activated. This mode is used by the boot code to load the map.bin
+//   file into memory.
+//
+// - In BOOT_VA mode, the _bdv_access() function uses a polling policy on
+//   IOC_STATUS register to detect transfer completion. This mode is used by
+//   the boot code to load the various .elf files into memory.
+//
+// - In KERNEL mode, the _bdv_access() function uses a descheduling strategy:
+//   The ISR executed when transfer completes should restart the calling task.
+//   There is no checking of user access right to the memory buffer. This mode
+//   must be used to access IOC, for an "open" system call.
+//
+// - In USER mode, the _bdv_access() function uses a descheduling strategy:
 //   The ISR executed when transfer completes should restart the calling task, 
 //   The user access right to the memory buffer must be checked.
 //   This mode must be used to access IOC, for a "read/write" system call.
 //
-// As the IOC component can be used by several programs running in parallel,
+// As the BDV component can be used by several programs running in parallel,
 // the _ioc_lock variable guaranties exclusive access to the device.  The
 // _bdv_read() and _bdv_write() functions use atomic LL/SC to get the lock.
-//
-// The IOMMU can be activated or not:
-// 
-// 1) When the IOMMU is used, a fixed size 2Mbytes vseg is allocated to 
-// the IOC peripheral, in the I/O virtual space, and the user buffer is
-// dynamically remapped in the IOMMU page table. The corresponding entry 
-// in the IOMMU PT1 is defined by the kernel _ioc_iommu_ix1 variable.
-// The number of pages to be unmapped is stored in the _ioc_npages variable.
-// The number of PT2 entries is dynamically computed and stored in the
-// kernel _ioc_iommu_npages variable. It cannot be larger than 512.
-// The user buffer is unmapped by the _ioc_completed() function when 
-// the transfer is completed.
-//
-// 2/ If the IOMMU is not used, we check that  the user buffer is mapped to a
-// contiguous physical buffer (this is generally true because the user space
-// page tables are statically constructed to use contiguous physical memory).
 //
 // Finally, the memory buffer must fulfill the following conditions: