[Cryptech-Commits] [sw/libhal] 01/05: Belatedly adjust hal_io_read/write to the new flat addressing architecture.

git at cryptech.is git at cryptech.is
Wed Jul 6 23:01:59 UTC 2016


This is an automated email from the git hooks/post-receive script.

paul at psgd.org pushed a commit to branch master
in repository sw/libhal.

commit 6a9c597bc9e6b9183355de8f99ce45611ea97a95
Author: Paul Selkirk <paul at psgd.org>
AuthorDate: Wed Jul 6 14:29:14 2016 -0400

    Belatedly adjust hal_io_read/write to the new flat addressing architecture.
---
 core.c       | 31 ++-----------------------------
 hal_io_eim.c | 11 ++---------
 hal_io_fmc.c | 11 ++---------
 3 files changed, 6 insertions(+), 47 deletions(-)

diff --git a/core.c b/core.c
index ffe61e6..0d7ed06 100644
--- a/core.c
+++ b/core.c
@@ -4,7 +4,7 @@
  * This module contains code to probe the FPGA for its installed cores.
  *
  * Author: Paul Selkirk, Rob Austein
- * Copyright (c) 2015, NORDUnet A/S All rights reserved.
+ * Copyright (c) 2015-2016, NORDUnet A/S All rights reserved.
  *
  * Redistribution and use in source and binary forms, with or without
  * modification, are permitted provided that the following conditions are
@@ -42,35 +42,8 @@
 #include "hal_internal.h"
 
 /*
- * Each Cryptech core has a set of 4-byte registers, which are accessed
- * through a 16-bit address. The address space is divided as follows:
- *   3 bits segment selector       | up to 8 segments
- *   5 bits core selector          | up to 32 cores/segment (see note below)
- *   8 bits register selector      | up to 256 registers/core (see modexp below)
- *
- * i.e, the address is structured as:
- * sss ccccc rrrrrrrr
- *
- * The I2C and UART communication channels use this 16-bit address format
- * directly in their read and write commands.
- *
- * The EIM communications channel translates this 16-bit address into a
- * 32-bit memory-mapped address in the range 0x08000000..807FFFF:
- * 00001000000000 sss 0 ccccc rrrrrrrr 00
- *
- * EIM, as implemented on the Novena, uses a 19-bit address space:
- *   Bits 18..16 are the semgent selector.
- *   Bits 15..10 are the core selector.
- *   Bits 9..2 are the register selector.
- *   Bits 1..0 are zero, because reads and writes are always word aligned.
- *
- * Note that EIM can support 64 cores per segment, but we sacrifice one bit
- * in order to map it into a 16-bit address space.
- */
-
-/*
  * Structure of our internal database is private, in case we want to
- * be change representation (array, tree, list of lists, whatever) at
+ * change representation (array, tree, list of lists, whatever) at
  * some later date without having to change the public API.
  */
 
diff --git a/hal_io_eim.c b/hal_io_eim.c
index 173490f..5824f5b 100644
--- a/hal_io_eim.c
+++ b/hal_io_eim.c
@@ -4,7 +4,7 @@
  * This module contains common code to talk to the FPGA over the EIM bus.
  *
  * Author: Paul Selkirk
- * Copyright (c) 2014-2015, NORDUnet A/S All rights reserved.
+ * Copyright (c) 2014-2016, NORDUnet A/S All rights reserved.
  *
  * Redistribution and use in source and binary forms, with or without
  * modification, are permitted provided that the following conditions are
@@ -60,17 +60,10 @@ static hal_error_t init(void)
 }
 
 /* translate cryptech register number to EIM address
- *
- * register number format:
- * 3 bits segment selector
- * 5 bits core selector (6 bits in native eim)
- * 8 bits register selector
- *
- * sss ccccc rrrrrrrr => 00001000000000 sss 0 ccccc rrrrrrrr 00
  */
 static hal_addr_t eim_offset(hal_addr_t offset)
 {
-  return EIM_BASE_ADDR + ((offset & ~0x1fff) << 3) + ((offset & 0x1fff) << 2);
+  return EIM_BASE_ADDR + (offset << 2);
 }
 
 void hal_io_set_debug(int onoff)
diff --git a/hal_io_fmc.c b/hal_io_fmc.c
index 5742e56..7aa4b19 100644
--- a/hal_io_fmc.c
+++ b/hal_io_fmc.c
@@ -4,7 +4,7 @@
  * This module contains common code to talk to the FPGA over the FMC bus.
  *
  * Author: Paul Selkirk
- * Copyright (c) 2014-2015, NORDUnet A/S All rights reserved.
+ * Copyright (c) 2014-2016, NORDUnet A/S All rights reserved.
  *
  * Redistribution and use in source and binary forms, with or without
  * modification, are permitted provided that the following conditions are
@@ -61,17 +61,10 @@ static hal_error_t init(void)
 }
 
 /* Translate cryptech register number to FMC address.
- *
- * register number format:
- * 3 bits segment selector
- * 5 bits core selector (6 bits in native eim)
- * 8 bits register selector
- *
- * sss ccccc rrrrrrrr => sss 0 ccccc rrrrrrrr 00
  */
 static hal_addr_t fmc_offset(hal_addr_t offset)
 {
-  return ((offset & ~0x1fff) << 3) + ((offset & 0x1fff) << 2);
+  return offset << 2;
 }
 
 void hal_io_set_debug(int onoff)



More information about the Commits mailing list