[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [PATCH 5/7] target/ppc: Rework ppc_radix64_walk_tree() for partition
From: |
Greg Kurz |
Subject: |
Re: [PATCH 5/7] target/ppc: Rework ppc_radix64_walk_tree() for partition-scoped translation |
Date: |
Mon, 30 Mar 2020 19:00:52 +0200 |
On Mon, 30 Mar 2020 11:49:44 +0200
Cédric Le Goater <address@hidden> wrote:
> The ppc_radix64_walk_tree() routine walks through the nested radix
> tables to look for a PTE.
>
> Split it two and introduce a new routine ppc_radix64_next_level()
Split it in two...
> which we will use for partition-scoped Radix translation when
> translating the process tree addresses.
>
> Signed-off-by: Suraj Jitindar Singh <address@hidden>
> Signed-off-by: Cédric Le Goater <address@hidden>
> ---
> target/ppc/mmu-radix64.c | 50 ++++++++++++++++++++++++++--------------
> 1 file changed, 33 insertions(+), 17 deletions(-)
>
> diff --git a/target/ppc/mmu-radix64.c b/target/ppc/mmu-radix64.c
> index b4e6abcd2d35..136498111f60 100644
> --- a/target/ppc/mmu-radix64.c
> +++ b/target/ppc/mmu-radix64.c
> @@ -162,44 +162,60 @@ static void ppc_radix64_set_rc(PowerPCCPU *cpu, int
> rwx, uint64_t pte,
> }
> }
>
> -static uint64_t ppc_radix64_walk_tree(PowerPCCPU *cpu, vaddr eaddr,
> - uint64_t base_addr, uint64_t nls,
> - hwaddr *raddr, int *psize,
> - int *fault_cause, hwaddr *pte_addr)
> +static uint64_t ppc_radix64_next_level(PowerPCCPU *cpu, vaddr eaddr,
> + uint64_t *pte_addr, uint64_t *nls,
> + int *psize, int *fault_cause)
> {
> CPUState *cs = CPU(cpu);
> uint64_t index, pde;
>
> - if (nls < 5) { /* Directory maps less than 2**5 entries */
> + if (*nls < 5) { /* Directory maps less than 2**5 entries */
> *fault_cause |= DSISR_R_BADCONFIG;
> return 0;
> }
>
I think this should stay in the caller...
> /* Read page <directory/table> entry from guest address space */
> - index = eaddr >> (*psize - nls); /* Shift */
> - index &= ((1UL << nls) - 1); /* Mask */
> - pde = ldq_phys(cs->as, base_addr + (index * sizeof(pde)));
> - if (!(pde & R_PTE_VALID)) { /* Invalid Entry */
> + pde = ldq_phys(cs->as, *pte_addr);
> + if (!(pde & R_PTE_VALID)) { /* Invalid Entry */
> *fault_cause |= DSISR_NOPTE;
> return 0;
> }
>
> - *psize -= nls;
> + *psize -= *nls;
> + if (!(pde & R_PTE_LEAF)) { /* Prepare for next iteration */
> + *nls = pde & R_PDE_NLS;
> + index = eaddr >> (*psize - *nls); /* Shift */
> + index &= ((1UL << *nls) - 1); /* Mask */
> + *pte_addr = (pde & R_PDE_NLB) + (index * sizeof(pde));
> + }
> + return pde;
> +}
> +
> +static uint64_t ppc_radix64_walk_tree(PowerPCCPU *cpu, vaddr eaddr,
> + uint64_t base_addr, uint64_t nls,
> + hwaddr *raddr, int *psize,
> + int *fault_cause, hwaddr *pte_addr)
> +{
> + uint64_t index, pde;
> +
> + index = eaddr >> (*psize - nls); /* Shift */
> + index &= ((1UL << nls) - 1); /* Mask */
> + *pte_addr = base_addr + (index * sizeof(pde));
> + do {
... here. So that we have a well established "bad config" path at
the root level, just like the current code has.
Since the ppc_radix64_walk_tree() now calls ppc_radix64_next_level()
in a loop instead of recursing, and since ppc_radix64_next_level()
returns the nls value for the next level, it really makes sense to
have this check in ppc_radix64_walk_tree() and maybe put an assert
in ppc_radix64_next_level().
> + pde = ppc_radix64_next_level(cpu, eaddr, pte_addr, &nls, psize,
> + fault_cause);
> + } while ((pde & R_PTE_VALID) && !(pde & R_PTE_LEAF));
>
> - /* Check if Leaf Entry -> Page Table Entry -> Stop the Search */
> - if (pde & R_PTE_LEAF) {
> + /* Did we find a valid leaf? */
> + if ((pde & R_PTE_VALID) && (pde & R_PTE_LEAF)) {
> uint64_t rpn = pde & R_PTE_RPN;
> uint64_t mask = (1UL << *psize) - 1;
>
> /* Or high bits of rpn and low bits to ea to form whole real addr */
> *raddr = (rpn & ~mask) | (eaddr & mask);
> - *pte_addr = base_addr + (index * sizeof(pde));
> - return pde;
> }
>
> - /* Next Level of Radix Tree */
> - return ppc_radix64_walk_tree(cpu, eaddr, pde & R_PDE_NLB, pde &
> R_PDE_NLS,
> - raddr, psize, fault_cause, pte_addr);
> + return pde;
> }
>
> static bool validate_pate(PowerPCCPU *cpu, uint64_t lpid, ppc_v3_pate_t
> *pate)
- Re: [PATCH 3/7] target/ppc: Assert if HV mode is set when running under a pseries machine, (continued)
- [PATCH 4/7] target/ppc: Introduce ppc_radix64_xlate() for Radix tree translation, Cédric Le Goater, 2020/03/30
- Re: [PATCH 4/7] target/ppc: Introduce ppc_radix64_xlate() for Radix tree translation, Greg Kurz, 2020/03/30
- Re: [PATCH 4/7] target/ppc: Introduce ppc_radix64_xlate() for Radix tree translation, Cédric Le Goater, 2020/03/30
- Re: [PATCH 4/7] target/ppc: Introduce ppc_radix64_xlate() for Radix tree translation, Cédric Le Goater, 2020/03/30
- Re: [PATCH 4/7] target/ppc: Introduce ppc_radix64_xlate() for Radix tree translation, Greg Kurz, 2020/03/30
- Re: [PATCH 4/7] target/ppc: Introduce ppc_radix64_xlate() for Radix tree translation, David Gibson, 2020/03/30
- Re: [PATCH 4/7] target/ppc: Introduce ppc_radix64_xlate() for Radix tree translation, Cédric Le Goater, 2020/03/31
[PATCH 5/7] target/ppc: Rework ppc_radix64_walk_tree() for partition-scoped translation, Cédric Le Goater, 2020/03/30
- Re: [PATCH 5/7] target/ppc: Rework ppc_radix64_walk_tree() for partition-scoped translation,
Greg Kurz <=
[PATCH 6/7] target/ppc: Extend ppc_radix64_check_prot() with a 'partition_scoped' bool, Cédric Le Goater, 2020/03/30
[PATCH 7/7] target/ppc: Add support for Radix partition-scoped translation, Cédric Le Goater, 2020/03/30