Commit f53e8676 authored by Robin Getz's avatar Robin Getz Committed by Bryan Wu

[Blackfin] arch: Add a note describing what is going on - no functional changes

Signed-off-by: default avatarRobin Getz <robin.getz@analog.com>
Signed-off-by: default avatarBryan Wu <bryan.wu@analog.com>
parent 326e96b9
......@@ -75,6 +75,15 @@ ENTRY(_cplb_mgr)
* from the configuration table.
*/
/* A multi-word instruction can cross a page boundary. This means the
* first part of the instruction can be in a valid page, but the
* second part is not, and hence generates the instruction miss.
* However, the fault address is for the start of the instruction,
* not the part that's in the bad page. Therefore, we have to check
* whether the fault address applies to a page that is already present
* in the table.
*/
P4.L = LO(ICPLB_FAULT_ADDR);
P4.H = HI(ICPLB_FAULT_ADDR);
......@@ -87,7 +96,7 @@ ENTRY(_cplb_mgr)
R4 = [P4]; /* Get faulting address*/
R6 = 64; /* Advance past the fault address, which*/
R6 = R6 + R4; /* we'll use if we find a match*/
R3 = ((16 << 8) | 2); /* Extract mask, bits 16 and 17.*/
R3 = ((16 << 8) | 2); /* Extract mask, two bits at posn 16 */
R5 = 0;
.Lisearch:
......@@ -125,7 +134,9 @@ ENTRY(_cplb_mgr)
P4.L = LO(IMEM_CONTROL);
P4.H = HI(IMEM_CONTROL);
/* disable cplbs */
/* Turn off CPLBs while we work, necessary according to HRM before
* modifying CPLB descriptors
*/
R5 = [P4]; /* Control Register*/
BITCLR(R5,ENICPLB_P);
CLI R1;
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment