summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorPeter Chubb <peterc@gelato.unsw.edu.au>2005-08-23 20:07:00 -0700
committerTony Luck <tony.luck@intel.com>2005-08-31 08:35:22 -0700
commit6cf07a8cc86a0b471466c7fe45892f7ef434015b (patch)
treeb1565448216586580ee34e6fa156353e8ec0540e
parent714d2dc14914f0f7bb008effe830c99eb47c75df (diff)
downloadop-kernel-dev-6cf07a8cc86a0b471466c7fe45892f7ef434015b.zip
op-kernel-dev-6cf07a8cc86a0b471466c7fe45892f7ef434015b.tar.gz
[IA64] Fix nasty VMLPT problem...
I've solved the problem I was having with the simulator and not booting Debian. The problem is that the number of bits for the virtual linear array short-format VHPT (Virtually mapped linear page table, VMLPT for short) is being tested incorrectly. There are two problems: 1. The PAL call that should tell the kernel the size of the virtual address space isn't implemented for the simulator, so the kernel uses the default 50. This is addressed separately in dc90e95f310f4f821c905b2aec8e9449bb3270fa 2. In arch/ia64/mm/init.c there's code to calcualte the size of the VMLPT based on the number of implemented virtual address bits and the page size. It checks to see if the VMLPT base address overlaps the top of the mapped region, but this check doesn't allow for the address space hole, and in fact will never trigger. Here's an alternative test and panic, that I think is more accurate. Signed-off-by: Peter Chubb <peterc@gelato.unsw.edu.au> Signed-off-by: Tony Luck <tony.luck@intel.com>
-rw-r--r--arch/ia64/mm/init.c15
1 files changed, 12 insertions, 3 deletions
diff --git a/arch/ia64/mm/init.c b/arch/ia64/mm/init.c
index 65f9958..1281c60 100644
--- a/arch/ia64/mm/init.c
+++ b/arch/ia64/mm/init.c
@@ -382,13 +382,22 @@ ia64_mmu_init (void *my_cpu_data)
if (impl_va_bits < 51 || impl_va_bits > 61)
panic("CPU has bogus IMPL_VA_MSB value of %lu!\n", impl_va_bits - 1);
+ /*
+ * mapped_space_bits - PAGE_SHIFT is the total number of ptes we need,
+ * which must fit into "vmlpt_bits - pte_bits" slots. Second half of
+ * the test makes sure that our mapped space doesn't overlap the
+ * unimplemented hole in the middle of the region.
+ */
+ if ((mapped_space_bits - PAGE_SHIFT > vmlpt_bits - pte_bits) ||
+ (mapped_space_bits > impl_va_bits - 1))
+ panic("Cannot build a big enough virtual-linear page table"
+ " to cover mapped address space.\n"
+ " Try using a smaller page size.\n");
+
/* place the VMLPT at the end of each page-table mapped region: */
pta = POW2(61) - POW2(vmlpt_bits);
- if (POW2(mapped_space_bits) >= pta)
- panic("mm/init: overlap between virtually mapped linear page table and "
- "mapped kernel space!");
/*
* Set the (virtually mapped linear) page table address. Bit
* 8 selects between the short and long format, bits 2-7 the
OpenPOWER on IntegriCloud