On Mon, Oct 16, 2017 at 2:34 PM, Aneesh Kumar K.V
<aneesh.ku...@linux.vnet.ibm.com> wrote:
> Michael Ellerman <m...@ellerman.id.au> writes:
>
>> From: Balbir Singh <bsinghar...@gmail.com>
>>
>> It would be nice to be able to dump page tables in a particular
>> context.
>>
>> eg: dumping vmalloc space:
>>
>> 0:mon> dv 0xd00037fffff00000
>> pgd @ 0xc0000000017c0000
>> pgdp @ 0xc0000000017c00d8 = 0x00000000f10b1000
>> pudp @ 0xc0000000f10b13f8 = 0x00000000f10d0000
>> pmdp @ 0xc0000000f10d1ff8 = 0x00000000f1102000
>> ptep @ 0xc0000000f1102780 = 0xc0000000f1ba018e
>> Maps physical address = 0x00000000f1ba0000
>> Flags = Accessed Dirty Read Write
>>
>> This patch does not replicate the complex code of dump_pagetable and
>> has no support for bolted linear mapping, thats why I've it's called
>> dump virtual page table support. The format of the PTE can be expanded
>> even further to add more useful information about the flags in the PTE
>> if required.
>>
>> Signed-off-by: Balbir Singh <bsinghar...@gmail.com>
>> [mpe: Bike shed the output format, show the pgdir]
>> Signed-off-by: Michael Ellerman <m...@ellerman.id.au>
>> ---
>> arch/powerpc/xmon/xmon.c | 111
>> +++++++++++++++++++++++++++++++++++++++++++++++
>> 1 file changed, 111 insertions(+)
>>
>> diff --git a/arch/powerpc/xmon/xmon.c b/arch/powerpc/xmon/xmon.c
>> index a41392dec822..d84bead0ac28 100644
>> --- a/arch/powerpc/xmon/xmon.c
>> +++ b/arch/powerpc/xmon/xmon.c
>> @@ -127,6 +127,7 @@ static void byterev(unsigned char *, int);
>> static void memex(void);
>> static int bsesc(void);
>> static void dump(void);
>> +static void show_pte(unsigned long);
>> static void prdump(unsigned long, long);
>> static int ppc_inst_dump(unsigned long, long, int);
>> static void dump_log_buf(void);
>> @@ -234,6 +235,7 @@ Commands:\n\
>> #endif
>> "\
>> dr dump stream of raw bytes\n\
>> + dv dump virtual address translation \n\
>> dt dump the tracing buffers (uses printk)\n\
>> dtc dump the tracing buffers for current CPU (uses printk)\n\
>> "
>> @@ -2639,6 +2641,9 @@ dump(void)
>> dump_log_buf();
>> } else if (c == 'o') {
>> dump_opal_msglog();
>> + } else if (c == 'v') {
>> + /* dump virtual to physical translation */
>> + show_pte(adrs);
>> } else if (c == 'r') {
>> scanhex(&ndump);
>> if (ndump == 0)
>> @@ -2972,6 +2977,112 @@ static void show_task(struct task_struct *tsk)
>> tsk->comm);
>> }
>>
>> +void format_pte(void *ptep, unsigned long pte)
>> +{
>> + printf("ptep @ 0x%016lx = 0x%016lx\n", (unsigned long)ptep, pte);;
>> + printf("Maps physical address = 0x%016lx\n", pte & PTE_RPN_MASK);
>> +
>> + printf("Flags = %s%s%s%s%s\n",
>> + (pte & _PAGE_ACCESSED) ? "Accessed " : "",
>> + (pte & _PAGE_DIRTY) ? "Dirty " : "",
>> + (pte & _PAGE_READ) ? "Read " : "",
>> + (pte & _PAGE_WRITE) ? "Write " : "",
>> + (pte & _PAGE_EXEC) ? "Exec " : "");
>> +}
>> +
>> +static void show_pte(unsigned long addr)
>> +{
>> + unsigned long tskv = 0;
>> + struct task_struct *tsk = NULL;
>> + struct mm_struct *mm;
>> + pgd_t *pgdp, *pgdir;
>> + pud_t *pudp;
>> + pmd_t *pmdp;
>> + pte_t *ptep;
>> +
>> + if (!scanhex(&tskv))
>> + mm = &init_mm;
>> + else
>> + tsk = (struct task_struct *)tskv;
>> +
>> + if (tsk == NULL)
>> + mm = &init_mm;
>> + else
>> + mm = tsk->active_mm;
>> +
>> + if (setjmp(bus_error_jmp) != 0) {
>> + catch_memory_errors = 0;
>> + printf("*** Error dumping pte for task %p\n", tsk);
>> + return;
>> + }
>> +
>> + catch_memory_errors = 1;
>> + sync();
>> +
>> + if (mm == &init_mm) {
>> + pgdp = pgd_offset_k(addr);
>> + pgdir = pgd_offset_k(0);
>> + } else {
>> + pgdp = pgd_offset(mm, addr);
>> + pgdir = pgd_offset(mm, 0);
>> + }
>> +
>> + if (pgd_none(*pgdp)) {
>> + printf("no linux page table for address\n");
>> + return;
>> + }
>> +
>> + printf("pgd @ 0x%016lx\n", pgdir);
>> +
>> + if (pgd_huge(*pgdp)) {
>> + format_pte(pgdp, pgd_val(*pgdp));
>> + return;
>> + }
>> + printf("pgdp @ 0x%016lx = 0x%016lx\n", pgdp, pgd_val(*pgdp));
>> +
>> + pudp = pud_offset(pgdp, addr);
>> +
>> + if (pud_none(*pudp)) {
>> + printf("No valid PUD\n");
>> + return;
>> + }
>> +
>> +#ifdef CONFIG_HUGETLB_PAGE
>> + if (pud_huge(*pudp)) {
>> + format_pte(pudp, pud_val(*pudp));
>> + return;
>> + }
>> +#endif
>
> For page table walking code we don't need to put #ifdef here.
OK.. I designed this for both, I'll take the #ifdef out for kernel page tables.
Also how
> should we handle address that we map at pmd level even if hugetlb page
> is disabled ? (kernel linear mapping). Also where do we handle THP
> mapping ? You may want to look at __find_linux_pte() to write a page
> table walking code. or better use the helper.
I wanted to avoid using __find_linux_pte in xmon(), it needs irq's disabled.
I found the interface a bit cumbersome. If I fix the #ifdef's and make the
walking code 64 bit only I think we should be OK? Do you agree?
Balbir Singh.
>
> -aneesh
>
