Clarify the sbiret.value

As per the calling convention a1 will have sbiret.value on ecall
return. However, the current specification describes an anlogous
C structure describing as long. Some of the SBI extension may require
an "unsigned long" data type instead of "long".

Relax the analgous C structure definition to allow that.

Fixes: #158

Signed-off-by: Atish Patra <[email protected]>

src/binary-encoding.adoc

@@ -23,7 +23,10 @@ of SBI extensions. The SBI specification follows the below calling convention.
 ----
     struct sbiret {
         long error;
-        long value;
+        union {
+            long value;
+            unsigned long uvalue;
+        };
     };
 ----
 

src/ext-debug-console.adoc

@@ -35,7 +35,7 @@ XLEN bits of the input memory physical base address.
 This is a non-blocking SBI call and it may do partial/no writes if
 the debug console is not able to accept more bytes.
 
-The number of bytes written is returned in `sbiret.value` and the
+The number of bytes written is returned in `sbiret.uvalue` and the
 possible error codes returned in `sbiret.error` are shown in
 <<table_debug_console_write_errors>> below.
 
@@ -75,7 +75,7 @@ This is a non-blocking SBI call and it will not write anything
 into the output memory if there are no bytes to be read in the
 debug console.
 
-The number of bytes read is returned in `sbiret.value` and the
+The number of bytes read is returned in `sbiret.uvalue` and the
 possible error codes returned in `sbiret.error` are shown in
 <<table_debug_console_read_errors>> below.
 
@@ -106,7 +106,7 @@ This is a blocking SBI call and it will only return after writing
 the specified byte to the debug console. It will also return, with
 SBI_ERR_FAILED, if there are I/O errors.
 
-The `sbiret.value` is set to zero and the possible error codes returned
+The `sbiret.uvalue` is set to zero and the possible error codes returned
 in `sbiret.error` are shown in <<table_debug_console_write_byte_errors>>
 below.
 

src/ext-mpxy.adoc

@@ -426,7 +426,7 @@ passed accordingly. The supervisor software may require multiple SBI calls to
 get the complete list of channel IDs depending on the `RETURNED` and
 `REMAINING` fields.
 
-The `sbiret.value` is always set to zero whereas the possible error codes
+The `sbiret.uvalue` is always set to zero whereas the possible error codes
 returned in `sbiret.error` are below.
 
 [#table_sbi_mpxy_get_channel_ids_errors]
@@ -598,15 +598,15 @@ specifications.
 
 Upon calling this function the SBI implementation MUST write the response
 message data at the offset `0x0` in the shared memory setup by the calling hart
-and the number of bytes written will be returned through `sbiret.value`.
+and the number of bytes written will be returned through `sbiret.uvalue`.
 The layout of data in case of both request and response is according to the
 respective message protocol specification message format.
 
 Upon success, this function: +
 1) Writes the message response data at offset `0x0` of the shared memory setup
 by the calling hart. +
 2) Returns `SBI_SUCCESS` in `sbiret.error`. +
-3) Returns message response data length in `sbiret.value`. +
+3) Returns message response data length in `sbiret.uvalue`. +
 
 This function is optional. If this function is implemented, the corresponding
 bit in the `CHANNEL_CAPABILITY` attribute is set to `1`.
@@ -769,7 +769,7 @@ SBI implementation at the offset `0x10` in the shared memory setup by the
 calling hart irrespective of events state data reporting. If events state data
 reporting is disabled or not supported, then the values in events state fields
 are undefined. The number of the bytes written to the shared memory will be
-returned through `sbiret.value` which is the number of bytes starting from
+returned through `sbiret.uvalue` which is the number of bytes starting from
 offset `0x10`.  The layout and encoding of notification events are defined by
 the message protocol specification associated with the message proxy channel
 (`channel_id`).