Title: [7885] trunk/arch/blackfin/kernel/setup.c: bug [#5662], don't leave 4KB unused before memory_end
- Revision
- 7885
- Author
- bhsong
- Date
- 2009-11-26 23:37:52 -0500 (Thu, 26 Nov 2009)
Log Message
bug [#5662], don't leave 4KB unused before memory_end
On one hand, according to the description of anomaly 05000310:
"Fetches at the boundary of either reserved memory or L1 Instruction cache memory (if instruction cache enabled) which is covered by a
valid CPLB cause a false Hardware Error (External Memory Addressing Error)."
Since it is impossible for our kernel to give a valid CPLB entry to reserved memory after physical memory, we have no neccessarity to place a 4KB hole.
On the other hand, even though I give a valid CPLB to 0x04000000, I have never triggerred this anomaly by fetching instructions and data at 0x03FFFFD0~0x04000000. The description is probably not right for all memory boundary cases.
Lastly, memory_end is usually not the end of valid memory due to DMA/MTD or anomaly.
Modified Paths
Diff
Modified: trunk/arch/blackfin/kernel/setup.c (7884 => 7885)
--- trunk/arch/blackfin/kernel/setup.c 2009-11-27 01:12:42 UTC (rev 7884)
+++ trunk/arch/blackfin/kernel/setup.c 2009-11-27 04:37:52 UTC (rev 7885)
@@ -601,11 +601,6 @@
page_mask_order = get_order(3 * page_mask_nelts * sizeof(long));
#endif
-#if !defined(CONFIG_MTD_UCLINUX)
- /*In case there is no valid CPLB behind memory_end make sure we don't get to close*/
- memory_end -= SIZE_4K;
-#endif
-
init_mm.start_code = (unsigned long)_stext;
init_mm.end_code = (unsigned long)_etext;
init_mm.end_data = (unsigned long)_edata;
_______________________________________________
Linux-kernel-commits mailing list
[email protected]
https://blackfin.uclinux.org/mailman/listinfo/linux-kernel-commits
