1:31 a.m.
I found Dave had alread done the first phase of future support for x86_64
5-level page tables(commit 307e7f35f510). when I asked him about the
state of this work, he gave me a more detailed answer and suggestion.
I follow his advice, and do the following job.
1. Refine the original logical:
1) Create some new common function for getting the offset of page table
2) Repace the PML4 and UPML with the common PGD:
machdep->machspec->pml4/upml ==> machdep->pdg
3) Using the PUD in x86_64
2. Add 5-level page tables support for x86_64_k/uvtop()
This patchset is the second phase of the work, As Dave said, we need to be
a manner of determining very early on whether the kernel page tables are
using 5-level and whether each user-space task is using 4- or 5-level page
tables. These will be done after this phase.
About test work:
I have tested this patchset with 4-level and 5-level paging table.
sadump/ Xen/ Old Linux / RHEL4 are not be tested.
Dou Liyang (5):
defs.h: Fix the PHYSICAL_PAGE_MASK macro
x86_64: Extract public page table mapping code
x86_64: Unify the page table parsing for 4-level mode
x86_64.c: Add the VMEMMAP support for 5 level page table
x86_64: Add 5-level paging support for x86_64_k/uvtop()
defs.h | 57 ++----
sadump.c | 9 +-
x86_64.c | 702 +++++++++++++++++++++++++++++++--------------------------------
3 files changed, 367 insertions(+), 401 deletions(-)
--
2.14.3
1:31 a.m.
New subject: [PATCH 1/5] defs.h: Fix the PHYSICAL_PAGE_MASK macro
Original crash defines the PHYSICAL_PAGE_MASK as
(~(PAGE_SIZE-1) & (__PHYSICAL_MASK << PAGE_SHIFT))
It moves left PAGE_SHIFT bits for safety, But As crash expands the physical
bits to 52, this will cause the conversion of phythcal address to virtual
address failed.
Remove the PAGE_SHIFT to fix this problem.
Signed-off-by: Dou Liyang <douly.fnst(a)cn.fujitsu.com>
---
Hi Dave,
The changelog may be not correct, I am not sure why does it move left
PAGE_SHIFT bits. Could you tell me the reason?
---
defs.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/defs.h b/defs.h
index dcd6c26..8082d6c 100644
--- a/defs.h
+++ b/defs.h
@@ -3403,7 +3403,7 @@ struct arm64_stackframe {
#define __VIRTUAL_MASK ((1UL << __VIRTUAL_MASK_SHIFT) - 1)
#define PAGE_SHIFT 12
#define PAGE_SIZE (1UL << PAGE_SHIFT)
-#define PHYSICAL_PAGE_MASK (~(PAGE_SIZE-1) & (__PHYSICAL_MASK <<
PAGE_SHIFT))
+#define PHYSICAL_PAGE_MASK (~(PAGE_SIZE-1) & __PHYSICAL_MASK )
#define _PAGE_BIT_NX 63
#define _PAGE_PRESENT 0x001
--
2.14.3
3 p.m.
New subject: [PATCH 1/5] defs.h: Fix the PHYSICAL_PAGE_MASK macro
----- Original Message -----
Original crash defines the PHYSICAL_PAGE_MASK as
(~(PAGE_SIZE-1) & (__PHYSICAL_MASK << PAGE_SHIFT))
It moves left PAGE_SHIFT bits for safety, But As crash expands the physical
bits to 52, this will cause the conversion of phythcal address to virtual
address failed.
Remove the PAGE_SHIFT to fix this problem.
Signed-off-by: Dou Liyang <douly.fnst(a)cn.fujitsu.com>
---
Hi Dave,
The changelog may be not correct, I am not sure why does it move left
PAGE_SHIFT bits. Could you tell me the reason?
I believe it's a (harmless so far) mistake. Good catch!
Dave
---
defs.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/defs.h b/defs.h
index dcd6c26..8082d6c 100644
--- a/defs.h
+++ b/defs.h
@@ -3403,7 +3403,7 @@ struct arm64_stackframe {
#define __VIRTUAL_MASK ((1UL << __VIRTUAL_MASK_SHIFT) - 1)
#define PAGE_SHIFT 12
#define PAGE_SIZE (1UL << PAGE_SHIFT)
-#define PHYSICAL_PAGE_MASK (~(PAGE_SIZE-1) & (__PHYSICAL_MASK <<
PAGE_SHIFT))
+#define PHYSICAL_PAGE_MASK (~(PAGE_SIZE-1) & __PHYSICAL_MASK )
#define _PAGE_BIT_NX 63
#define _PAGE_PRESENT 0x001
--
2.14.3
1:31 a.m.
New subject: [PATCH 2/5] x86_64: Extract public page table mapping code
Currently, page table mapping code is scattered everywhere. So Extract public
page table mapping code, Prepare for 5-level page
No functionality change
Signed-off-by: Dou Liyang <douly.fnst(a)cn.fujitsu.com>
---
x86_64.c | 410 ++++++++++++++++++++++++++++++---------------------------------
1 file changed, 193 insertions(+), 217 deletions(-)
diff --git a/x86_64.c b/x86_64.c
index d8fade4..7134b4b 100644
--- a/x86_64.c
+++ b/x86_64.c
@@ -78,6 +78,11 @@ static void x86_64_calc_phys_base(void);
static int x86_64_is_module_addr(ulong);
static int x86_64_is_kvaddr(ulong);
static int x86_64_is_uvaddr(ulong, struct task_context *);
+static ulong *x86_64_kpgd_offset(ulong, int, int);
+static ulong x86_64_upgd_offset(struct task_context *, ulong, int, int);
+static ulong x86_64_pud_offset(ulong, ulong, int, int);
+static ulong x86_64_pmd_offset(ulong, ulong, int, int);
+static ulong x86_64_pte_offset(ulong, ulong, int, int);
void x86_64_compiler_warning_stub(void);
static void x86_64_init_kernel_pgd(void);
static void x86_64_cpu_pda_init(void);
@@ -1557,6 +1562,153 @@ x86_64_is_uvaddr(ulong addr, struct task_context *tc)
return (addr < USERSPACE_TOP);
}
+/*
+ * Find the kernel pgd entry..
+ * pgd = pgd_offset_k(addr);
+ */
+static ulong *
+x86_64_kpgd_offset(ulong kvaddr, int verbose , int IS_XEN)
+{
+ ulong *pgd;
+
+ FILL_PML4();
+ pgd = ((ulong *)machdep->machspec->pml4) + pml4_index(kvaddr);
+ if (verbose) {
+ fprintf(fp, "PGD DIRECTORY: %lx\n", vt->kernel_pgd[0]);
+ if(IS_XEN)
+ fprintf(fp, "PAGE DIRECTORY: %lx [machine]\n", *pgd);
+ else
+ fprintf(fp, "PAGE DIRECTORY: %lx\n", *pgd);
+ }
+
+ return pgd;
+}
+
+/*
+ * Find the user pgd entry..
+ * pgd = pgd_offset(mm, address);
+ */
+static ulong
+x86_64_upgd_offset(struct task_context *tc, ulong uvaddr, int verbose , int IS_XEN)
+{
+ ulong *pgd;
+ ulong pgd_paddr;
+ ulong pgd_pte;
+
+ if (task_mm(tc->task, TRUE))
+ pgd = ULONG_PTR(tt->mm_struct + OFFSET(mm_struct_pgd));
+ else
+ readmem(tc->mm_struct + OFFSET(mm_struct_pgd), KVADDR, &pgd,
+ sizeof(long), "mm_struct pgd", FAULT_ON_ERROR);
+
+ pgd_paddr = x86_64_VTOP((ulong)pgd);
+ FILL_UPML(pgd_paddr, PHYSADDR, PAGESIZE());
+ pgd = ((ulong *)pgd_paddr) + pml4_index(uvaddr);
+ pgd_pte = ULONG(machdep->machspec->pml4 + PAGEOFFSET(pgd));
+ if (verbose) {
+ if(IS_XEN)
+ fprintf(fp, " PGD: %lx => %lx [machine]\n", (ulong)pgd, pgd_pte);
+ else
+ fprintf(fp, " PGD: %lx => %lx\n", (ulong)pgd, pgd_pte);
+ }
+
+ return pgd_pte;
+}
+
+/*
+ * Find an entry in the third-level page table..
+ * pud = pud_offset(pgd, address);
+ */
+static ulong
+x86_64_pud_offset(ulong pgd_pte, ulong vaddr, int verbose , int IS_XEN)
+{
+ ulong *pud;
+ ulong pud_paddr;
+ ulong pud_pte;
+
+ pud_paddr = pgd_pte & PHYSICAL_PAGE_MASK;
+
+ if(IS_XEN) {
+ pud_paddr = xen_m2p(pud_paddr);
+ if (verbose)
+ fprintf(fp, " PGD: %lx\n", pud_paddr);
+ }
+
+ FILL_PUD(pud_paddr, PHYSADDR, PAGESIZE());
+ pud = ((ulong *)pud_paddr) + pgd_index(vaddr);
+ pud_pte = ULONG(machdep->pud + PAGEOFFSET(pud));
+ if (verbose) {
+ if(IS_XEN)
+ fprintf(fp, " PUD: %lx => %lx [machine]\n", (ulong)pud, pud_pte);
+ else
+ fprintf(fp, " PUD: %lx => %lx\n", (ulong)pud, pud_pte);
+ }
+
+ return pud_pte;
+}
+
+/*
+ * Find an entry in the middle page table..
+ * pmd = pmd_offset(pud, address);
+ */
+static ulong
+x86_64_pmd_offset(ulong pud_pte, ulong vaddr, int verbose , int IS_XEN)
+{
+ ulong *pmd;
+ ulong pmd_paddr;
+ ulong pmd_pte;
+
+ pmd_paddr = pud_pte & PHYSICAL_PAGE_MASK;
+
+ if(IS_XEN) {
+ pmd_paddr = xen_m2p(pmd_paddr);
+ if (verbose)
+ fprintf(fp, " PUD: %lx\n", pmd_paddr);
+ }
+
+ FILL_PMD(pmd_paddr, PHYSADDR, PAGESIZE());
+ pmd = ((ulong *)pmd_paddr) + pmd_index(vaddr);
+ pmd_pte = ULONG(machdep->pmd + PAGEOFFSET(pmd));
+ if (verbose) {
+ if(IS_XEN)
+ fprintf(fp, " PMD: %lx => %lx [machine]\n", (ulong)pmd, pmd_pte);
+ else
+ fprintf(fp, " PMD: %lx => %lx\n", (ulong)pmd, pmd_pte);
+ }
+ return pmd_pte;
+}
+
+/*
+ * Find an entry in the pet page table..
+ * pmd = pmd_offset(pud, address);
+ */
+static ulong
+x86_64_pte_offset(ulong pmd_pte, ulong vaddr, int verbose , int IS_XEN)
+{
+ ulong *ptep;
+ ulong pte_paddr;
+ ulong pte;
+
+ pte_paddr = pmd_pte & PHYSICAL_PAGE_MASK;
+
+ if(IS_XEN) {
+ pte_paddr = xen_m2p(pte_paddr);
+ if (verbose)
+ fprintf(fp, " PMD: %lx\n", pte_paddr);
+ }
+
+ FILL_PTBL(pte_paddr, PHYSADDR, PAGESIZE());
+ ptep = ((ulong *)pte_paddr) + pte_index(vaddr);
+ pte = ULONG(machdep->ptbl + PAGEOFFSET(ptep));
+ if (verbose) {
+ if(IS_XEN)
+ fprintf(fp, " PTE: %lx => %lx [machine]\n", (ulong)ptep, pte);
+ else
+ fprintf(fp, " PTE: %lx => %lx\n", (ulong)ptep, pte);
+ }
+
+ return pte;
+}
/*
* Translates a user virtual address to its physical address. cmd_vtop()
@@ -1570,18 +1722,9 @@ x86_64_is_uvaddr(ulong addr, struct task_context *tc)
static int
x86_64_uvtop_level4(struct task_context *tc, ulong uvaddr, physaddr_t *paddr, int
verbose)
{
- ulong mm;
- ulong *pml;
- ulong pml_paddr;
- ulong pml_pte;
- ulong *pgd;
- ulong pgd_paddr;
ulong pgd_pte;
- ulong *pmd;
- ulong pmd_paddr;
+ ulong pud_pte;
ulong pmd_pte;
- ulong *ptep;
- ulong pte_paddr;
ulong pte;
physaddr_t physpage;
@@ -1596,43 +1739,25 @@ x86_64_uvtop_level4(struct task_context *tc, ulong uvaddr,
physaddr_t *paddr, in
if ((machdep->flags & VM_5LEVEL) && x86_64_task_uses_5level(tc))
return x86_64_uvtop_5level(tc, uvaddr, paddr, verbose);
- if ((mm = task_mm(tc->task, TRUE)))
- pml = ULONG_PTR(tt->mm_struct + OFFSET(mm_struct_pgd));
- else
- readmem(tc->mm_struct + OFFSET(mm_struct_pgd), KVADDR, &pml,
- sizeof(long), "mm_struct pgd", FAULT_ON_ERROR);
-
- pml_paddr = x86_64_VTOP((ulong)pml);
- FILL_UPML(pml_paddr, PHYSADDR, PAGESIZE());
- pml = ((ulong *)pml_paddr) + pml4_index(uvaddr);
- pml_pte = ULONG(machdep->machspec->upml + PAGEOFFSET(pml));
- if (verbose)
- fprintf(fp, " PML: %lx => %lx\n", (ulong)pml, pml_pte);
- if (!(pml_pte & _PAGE_PRESENT))
+ pgd_pte = x86_64_upgd_offset(tc, uvaddr, verbose, FALSE);
+ if (!(pgd_pte & _PAGE_PRESENT))
goto no_upage;
- pgd_paddr = pml_pte & PHYSICAL_PAGE_MASK;
- FILL_PGD(pgd_paddr, PHYSADDR, PAGESIZE());
- pgd = ((ulong *)pgd_paddr) + pgd_index(uvaddr);
- pgd_pte = ULONG(machdep->pgd + PAGEOFFSET(pgd));
- if (verbose)
- fprintf(fp, " PUD: %lx => %lx\n", (ulong)pgd, pgd_pte);
- if (!(pgd_pte & _PAGE_PRESENT))
+ /*
+ * pud = pud_offset(pgd, address);
+ */
+ pud_pte = x86_64_pud_offset(pgd_pte, uvaddr, verbose, FALSE);
+ if (!(pud_pte & _PAGE_PRESENT))
goto no_upage;
/*
- * pmd = pmd_offset(pgd, address);
+ * pmd = pmd_offset(pud, address);
*/
- pmd_paddr = pgd_pte & PHYSICAL_PAGE_MASK;
- FILL_PMD(pmd_paddr, PHYSADDR, PAGESIZE());
- pmd = ((ulong *)pmd_paddr) + pmd_index(uvaddr);
- pmd_pte = ULONG(machdep->pmd + PAGEOFFSET(pmd));
- if (verbose)
- fprintf(fp, " PMD: %lx => %lx\n", (ulong)pmd, pmd_pte);
+ pmd_pte = x86_64_pmd_offset(pud_pte, uvaddr, verbose, FALSE);
if (!(pmd_pte & (_PAGE_PRESENT | _PAGE_PROTNONE)))
goto no_upage;
if (pmd_pte & _PAGE_PSE) {
- if (verbose) {
+ if (verbose) {
fprintf(fp, " PAGE: %lx (2MB)\n\n",
PAGEBASE(pmd_pte) & PHYSICAL_PAGE_MASK);
x86_64_translate_pte(pmd_pte, 0, 0);
@@ -1648,13 +1773,7 @@ x86_64_uvtop_level4(struct task_context *tc, ulong uvaddr,
physaddr_t *paddr, in
* ptep = pte_offset_map(pmd, address);
* pte = *ptep;
*/
- pte_paddr = pmd_pte & PHYSICAL_PAGE_MASK;
- FILL_PTBL(pte_paddr, PHYSADDR, PAGESIZE());
- ptep = ((ulong *)pte_paddr) + pte_index(uvaddr);
- pte = ULONG(machdep->ptbl + PAGEOFFSET(ptep));
- if (verbose)
- fprintf(fp, " PTE: %lx => %lx\n", (ulong)ptep, pte);
-
+ pte = x86_64_pte_offset(pmd_pte, uvaddr, verbose, FALSE);
if (!(pte & (_PAGE_PRESENT | _PAGE_PROTNONE))) {
*paddr = pte;
@@ -1696,19 +1815,10 @@ x86_64_uvtop_5level(struct task_context *tc, ulong uvaddr,
physaddr_t *paddr, in
static int
x86_64_uvtop_level4_xen_wpt(struct task_context *tc, ulong uvaddr, physaddr_t *paddr, int
verbose)
{
- ulong mm;
- ulong *pml;
- ulong pml_paddr;
- ulong pml_pte;
- ulong *pgd;
- ulong pgd_paddr;
ulong pgd_pte;
- ulong *pmd;
- ulong pmd_paddr;
+ ulong pud_pte;
ulong pmd_pte;
ulong pseudo_pmd_pte;
- ulong *ptep;
- ulong pte_paddr;
ulong pte;
ulong pseudo_pte;
physaddr_t physpage;
@@ -1722,45 +1832,18 @@ x86_64_uvtop_level4_xen_wpt(struct task_context *tc, ulong uvaddr,
physaddr_t *p
if (IS_KVADDR(uvaddr))
return x86_64_kvtop(tc, uvaddr, paddr, verbose);
- if ((mm = task_mm(tc->task, TRUE)))
- pml = ULONG_PTR(tt->mm_struct + OFFSET(mm_struct_pgd));
- else
- readmem(tc->mm_struct + OFFSET(mm_struct_pgd), KVADDR, &pml,
- sizeof(long), "mm_struct pgd", FAULT_ON_ERROR);
-
- pml_paddr = x86_64_VTOP((ulong)pml);
- FILL_UPML(pml_paddr, PHYSADDR, PAGESIZE());
- pml = ((ulong *)pml_paddr) + pml4_index(uvaddr);
- pml_pte = ULONG(machdep->machspec->upml + PAGEOFFSET(pml));
- if (verbose)
- fprintf(fp, " PML: %lx => %lx [machine]\n", (ulong)pml, pml_pte);
- if (!(pml_pte & _PAGE_PRESENT))
+ pgd_pte = x86_64_upgd_offset(tc, uvaddr, verbose, TRUE);
+ if (!(pgd_pte & _PAGE_PRESENT))
goto no_upage;
- pgd_paddr = pml_pte & PHYSICAL_PAGE_MASK;
- pgd_paddr = xen_m2p(pgd_paddr);
- if (verbose)
- fprintf(fp, " PML: %lx\n", pgd_paddr);
- FILL_PGD(pgd_paddr, PHYSADDR, PAGESIZE());
- pgd = ((ulong *)pgd_paddr) + pgd_index(uvaddr);
- pgd_pte = ULONG(machdep->pgd + PAGEOFFSET(pgd));
- if (verbose)
- fprintf(fp, " PUD: %lx => %lx [machine]\n", (ulong)pgd,
pgd_pte);
- if (!(pgd_pte & _PAGE_PRESENT))
+ pud_pte = x86_64_pud_offset(pgd_pte, uvaddr, verbose, TRUE);
+ if (!(pud_pte & _PAGE_PRESENT))
goto no_upage;
/*
- * pmd = pmd_offset(pgd, address);
+ * pmd = pmd_offset(pud, address);
*/
- pmd_paddr = pgd_pte & PHYSICAL_PAGE_MASK;
- pmd_paddr = xen_m2p(pmd_paddr);
- if (verbose)
- fprintf(fp, " PUD: %lx\n", pmd_paddr);
- FILL_PMD(pmd_paddr, PHYSADDR, PAGESIZE());
- pmd = ((ulong *)pmd_paddr) + pmd_index(uvaddr);
- pmd_pte = ULONG(machdep->pmd + PAGEOFFSET(pmd));
- if (verbose)
- fprintf(fp, " PMD: %lx => %lx [machine]\n", (ulong)pmd,
pmd_pte);
+ pmd_pte = x86_64_pmd_offset(pud_pte, uvaddr, verbose, TRUE);
if (!(pmd_pte & _PAGE_PRESENT))
goto no_upage;
if (pmd_pte & _PAGE_PSE) {
@@ -1799,15 +1882,7 @@ x86_64_uvtop_level4_xen_wpt(struct task_context *tc, ulong uvaddr,
physaddr_t *p
* ptep = pte_offset_map(pmd, address);
* pte = *ptep;
*/
- pte_paddr = pmd_pte & PHYSICAL_PAGE_MASK;
- pte_paddr = xen_m2p(pte_paddr);
- if (verbose)
- fprintf(fp, " PMD: %lx\n", pte_paddr);
- FILL_PTBL(pte_paddr, PHYSADDR, PAGESIZE());
- ptep = ((ulong *)pte_paddr) + pte_index(uvaddr);
- pte = ULONG(machdep->ptbl + PAGEOFFSET(ptep));
- if (verbose)
- fprintf(fp, " PTE: %lx => %lx [machine]\n", (ulong)ptep, pte);
+ pte = x86_64_pte_offset(pmd_pte, uvaddr, verbose, TRUE);
if (!(pte & (_PAGE_PRESENT))) {
*paddr = pte;
@@ -1817,7 +1892,7 @@ x86_64_uvtop_level4_xen_wpt(struct task_context *tc, ulong uvaddr,
physaddr_t *p
}
goto no_upage;
}
-
+
pseudo_pte = xen_m2p(pte & PHYSICAL_PAGE_MASK);
if (verbose)
fprintf(fp, " PTE: %lx\n", pseudo_pte + PAGEOFFSET(pte));
@@ -1846,12 +1921,8 @@ x86_64_uvtop_level4_rhel4_xen_wpt(struct task_context *tc, ulong
uvaddr, physadd
ulong *pgd;
ulong pgd_paddr;
ulong pgd_pte;
- ulong *pmd;
- ulong pmd_paddr;
ulong pmd_pte;
ulong pseudo_pmd_pte;
- ulong *ptep;
- ulong pte_paddr;
ulong pte;
ulong pseudo_pte;
physaddr_t physpage;
@@ -1883,15 +1954,7 @@ x86_64_uvtop_level4_rhel4_xen_wpt(struct task_context *tc, ulong
uvaddr, physadd
/*
* pmd = pmd_offset(pgd, address);
*/
- pmd_paddr = pgd_pte & PHYSICAL_PAGE_MASK;
- pmd_paddr = xen_m2p(pmd_paddr);
- if (verbose)
- fprintf(fp, " PGD: %lx\n", pmd_paddr);
- FILL_PMD(pmd_paddr, PHYSADDR, PAGESIZE());
- pmd = ((ulong *)pmd_paddr) + pmd_index(uvaddr);
- pmd_pte = ULONG(machdep->pmd + PAGEOFFSET(pmd));
- if (verbose)
- fprintf(fp, " PMD: %lx => %lx [machine]\n", (ulong)pmd,
pmd_pte);
+ pmd_pte = x86_64_pmd_offset(pgd_pte, uvaddr, verbose, TRUE);
if (!(pmd_pte & _PAGE_PRESENT))
goto no_upage;
if (pmd_pte & _PAGE_PSE) {
@@ -1930,15 +1993,7 @@ x86_64_uvtop_level4_rhel4_xen_wpt(struct task_context *tc, ulong
uvaddr, physadd
* ptep = pte_offset_map(pmd, address);
* pte = *ptep;
*/
- pte_paddr = pmd_pte & PHYSICAL_PAGE_MASK;
- pte_paddr = xen_m2p(pte_paddr);
- if (verbose)
- fprintf(fp, " PMD: %lx\n", pte_paddr);
- FILL_PTBL(pte_paddr, PHYSADDR, PAGESIZE());
- ptep = ((ulong *)pte_paddr) + pte_index(uvaddr);
- pte = ULONG(machdep->ptbl + PAGEOFFSET(ptep));
- if (verbose)
- fprintf(fp, " PTE: %lx => %lx [machine]\n", (ulong)ptep, pte);
+ pte = x86_64_pte_offset(pmd_pte, uvaddr, verbose, TRUE);
if (!(pte & (_PAGE_PRESENT))) {
*paddr = pte;
@@ -1977,11 +2032,7 @@ x86_64_uvtop(struct task_context *tc, ulong uvaddr, physaddr_t
*paddr, int verbo
ulong *pgd;
ulong pgd_paddr;
ulong pgd_pte;
- ulong *pmd;
- ulong pmd_paddr;
ulong pmd_pte;
- ulong *ptep;
- ulong pte_paddr;
ulong pte;
physaddr_t physpage;
@@ -2014,12 +2065,7 @@ x86_64_uvtop(struct task_context *tc, ulong uvaddr, physaddr_t
*paddr, int verbo
/*
* pmd = pmd_offset(pgd, address);
*/
- pmd_paddr = pgd_pte & PHYSICAL_PAGE_MASK;
- FILL_PMD(pmd_paddr, PHYSADDR, PAGESIZE());
- pmd = ((ulong *)pmd_paddr) + pmd_index(uvaddr);
- pmd_pte = ULONG(machdep->pmd + PAGEOFFSET(pmd));
- if (verbose)
- fprintf(fp, " PMD: %lx => %lx\n", (ulong)pmd, pmd_pte);
+ pmd_pte = x86_64_pmd_offset(pgd_pte, uvaddr, verbose, FALSE);
if (!(pmd_pte & _PAGE_PRESENT))
goto no_upage;
if (pmd_pte & _PAGE_PSE) {
@@ -2039,12 +2085,7 @@ x86_64_uvtop(struct task_context *tc, ulong uvaddr, physaddr_t
*paddr, int verbo
* ptep = pte_offset_map(pmd, address);
* pte = *ptep;
*/
- pte_paddr = pmd_pte & PHYSICAL_PAGE_MASK;
- FILL_PTBL(pte_paddr, PHYSADDR, PAGESIZE());
- ptep = ((ulong *)pte_paddr) + pte_index(uvaddr);
- pte = ULONG(machdep->ptbl + PAGEOFFSET(ptep));
- if (verbose)
- fprintf(fp, " PTE: %lx => %lx\n", (ulong)ptep, pte);
+ pte = x86_64_pte_offset(pmd_pte, uvaddr, verbose, FALSE);
if (!(pte & (_PAGE_PRESENT))) {
*paddr = pte;
@@ -2070,7 +2111,6 @@ no_upage:
return FALSE;
}
-
/*
* Translates a kernel virtual address to its physical address. cmd_vtop()
* sets the verbose flag so that the pte translation gets displayed; all
@@ -2080,14 +2120,8 @@ static int
x86_64_kvtop(struct task_context *tc, ulong kvaddr, physaddr_t *paddr, int verbose)
{
ulong *pml4;
- ulong *pgd;
- ulong pgd_paddr;
- ulong pgd_pte;
- ulong *pmd;
- ulong pmd_paddr;
+ ulong pud_pte;
ulong pmd_pte;
- ulong *ptep;
- ulong pte_paddr;
ulong pte;
physaddr_t physpage;
@@ -2098,12 +2132,7 @@ x86_64_kvtop(struct task_context *tc, ulong kvaddr, physaddr_t
*paddr, int verbo
* stage phys_base is not initialized yet and x86_64_VTOP()
* does not work. Jump to the code of pagetable translation.
*/
- FILL_PML4();
- pml4 = ((ulong *)machdep->machspec->pml4) + pml4_index(kvaddr);
- if (verbose) {
- fprintf(fp, "PML4 DIRECTORY: %lx\n", vt->kernel_pgd[0]);
- fprintf(fp, "PAGE DIRECTORY: %lx\n", *pml4);
- }
+ pgd = x86_64_kpgd_offset(kvaddr, verbose, FALSE);
goto start_vtop_with_pagetable;
}
@@ -2146,35 +2175,21 @@ x86_64_kvtop(struct task_context *tc, ulong kvaddr, physaddr_t
*paddr, int verbo
/*
* pgd = pgd_offset_k(addr);
*/
- FILL_PML4();
- pml4 = ((ulong *)machdep->machspec->pml4) + pml4_index(kvaddr);
- if (verbose) {
- fprintf(fp, "PML4 DIRECTORY: %lx\n", vt->kernel_pgd[0]);
- fprintf(fp, "PAGE DIRECTORY: %lx\n", *pml4);
- }
+ pgd = x86_64_kpgd_offset(kvaddr, verbose, FALSE);
}
start_vtop_with_pagetable:
if (!(*pml4 & _PAGE_PRESENT))
goto no_kpage;
- pgd_paddr = (*pml4) & PHYSICAL_PAGE_MASK;
- FILL_PGD(pgd_paddr, PHYSADDR, PAGESIZE());
- pgd = ((ulong *)pgd_paddr) + pgd_index(kvaddr);
- pgd_pte = ULONG(machdep->pgd + PAGEOFFSET(pgd));
- if (verbose)
- fprintf(fp, " PUD: %lx => %lx\n", (ulong)pgd, pgd_pte);
- if (!(pgd_pte & _PAGE_PRESENT))
+
+ pud_pte = x86_64_pud_offset(*pml4, kvaddr, verbose, FALSE);
+ if (!(pud_pte & _PAGE_PRESENT))
goto no_kpage;
/*
- * pmd = pmd_offset(pgd, addr);
+ * pmd = pmd_offset(pud, address);
*/
- pmd_paddr = pgd_pte & PHYSICAL_PAGE_MASK;
- FILL_PMD(pmd_paddr, PHYSADDR, PAGESIZE());
- pmd = ((ulong *)pmd_paddr) + pmd_index(kvaddr);
- pmd_pte = ULONG(machdep->pmd + PAGEOFFSET(pmd));
- if (verbose)
- fprintf(fp, " PMD: %lx => %lx\n", (ulong)pmd, pmd_pte);
+ pmd_pte = x86_64_pmd_offset(pud_pte, kvaddr, verbose, FALSE);
if (!(pmd_pte & _PAGE_PRESENT))
goto no_kpage;
if (pmd_pte & _PAGE_PSE) {
@@ -2194,12 +2209,7 @@ start_vtop_with_pagetable:
* ptep = pte_offset_map(pmd, addr);
* pte = *ptep;
*/
- pte_paddr = pmd_pte & PHYSICAL_PAGE_MASK;
- FILL_PTBL(pte_paddr, PHYSADDR, PAGESIZE());
- ptep = ((ulong *)pte_paddr) + pte_index(kvaddr);
- pte = ULONG(machdep->ptbl + PAGEOFFSET(ptep));
- if (verbose)
- fprintf(fp, " PTE: %lx => %lx\n", (ulong)ptep, pte);
+ pte = x86_64_pte_offset(pmd_pte, kvaddr, verbose, FALSE);
if (!(pte & (_PAGE_PRESENT))) {
if (pte && verbose) {
fprintf(fp, "\n");
@@ -2233,15 +2243,9 @@ static int
x86_64_kvtop_xen_wpt(struct task_context *tc, ulong kvaddr, physaddr_t *paddr, int
verbose)
{
ulong *pml4;
- ulong *pgd;
- ulong pgd_paddr;
- ulong pgd_pte;
- ulong *pmd;
- ulong pmd_paddr;
+ ulong pud_pte;
ulong pmd_pte;
ulong pseudo_pmd_pte;
- ulong *ptep;
- ulong pte_paddr;
ulong pte;
ulong pseudo_pte;
physaddr_t physpage;
@@ -2250,38 +2254,18 @@ x86_64_kvtop_xen_wpt(struct task_context *tc, ulong kvaddr,
physaddr_t *paddr, i
/*
* pgd = pgd_offset_k(addr);
*/
- FILL_PML4();
- pml4 = ((ulong *)machdep->machspec->pml4) + pml4_index(kvaddr);
- if (verbose) {
- fprintf(fp, "PML4 DIRECTORY: %lx\n", vt->kernel_pgd[0]);
- fprintf(fp, "PAGE DIRECTORY: %lx [machine]\n", *pml4);
- }
+ pml4 = x86_64_kpgd_offset(kvaddr, verbose, TRUE);
if (!(*pml4 & _PAGE_PRESENT))
goto no_kpage;
- pgd_paddr = (*pml4) & PHYSICAL_PAGE_MASK;
- pgd_paddr = xen_m2p(pgd_paddr);
- if (verbose)
- fprintf(fp, "PAGE DIRECTORY: %lx\n", pgd_paddr);
- FILL_PGD(pgd_paddr, PHYSADDR, PAGESIZE());
- pgd = ((ulong *)pgd_paddr) + pgd_index(kvaddr);
- pgd_pte = ULONG(machdep->pgd + PAGEOFFSET(pgd));
- if (verbose)
- fprintf(fp, " PUD: %lx => %lx [machine]\n", (ulong)pgd,
pgd_pte);
- if (!(pgd_pte & _PAGE_PRESENT))
+
+ pud_pte = x86_64_pud_offset(*pml4, kvaddr, verbose, TRUE);
+ if (!(pud_pte & _PAGE_PRESENT))
goto no_kpage;
/*
* pmd = pmd_offset(pgd, addr);
*/
- pmd_paddr = pgd_pte & PHYSICAL_PAGE_MASK;
- pmd_paddr = xen_m2p(pmd_paddr);
- if (verbose)
- fprintf(fp, " PUD: %lx\n", pmd_paddr);
- FILL_PMD(pmd_paddr, PHYSADDR, PAGESIZE());
- pmd = ((ulong *)pmd_paddr) + pmd_index(kvaddr);
- pmd_pte = ULONG(machdep->pmd + PAGEOFFSET(pmd));
- if (verbose)
- fprintf(fp, " PMD: %lx => %lx [machine]\n", (ulong)pmd,
pmd_pte);
+ pmd_pte = x86_64_pmd_offset(pud_pte, kvaddr, verbose, TRUE);
if (!(pmd_pte & _PAGE_PRESENT))
goto no_kpage;
if (pmd_pte & _PAGE_PSE) {
@@ -2320,15 +2304,7 @@ x86_64_kvtop_xen_wpt(struct task_context *tc, ulong kvaddr,
physaddr_t *paddr, i
* ptep = pte_offset_map(pmd, addr);
* pte = *ptep;
*/
- pte_paddr = pmd_pte & PHYSICAL_PAGE_MASK;
- pte_paddr = xen_m2p(pte_paddr);
- if (verbose)
- fprintf(fp, " PMD: %lx\n", pte_paddr);
- FILL_PTBL(pte_paddr, PHYSADDR, PAGESIZE());
- ptep = ((ulong *)pte_paddr) + pte_index(kvaddr);
- pte = ULONG(machdep->ptbl + PAGEOFFSET(ptep));
- if (verbose)
- fprintf(fp, " PTE: %lx => %lx [machine]\n", (ulong)ptep,
pte);
+ pte = x86_64_pte_offset(pmd_pte, kvaddr, verbose, TRUE);
if (!(pte & (_PAGE_PRESENT))) {
if (pte && verbose) {
fprintf(fp, "\n");
--
2.14.3
1:31 a.m.
New subject: [PATCH 3/5] x86_64: Unify the page table parsing for 4-level mode
Repace the PML4 and UPML with the common PGD to make page table uniform
for supporting 5 level mode.
Signed-off-by: Dou Liyang <douly.fnst(a)cn.fujitsu.com>
---
defs.h | 55 ++++++---------
sadump.c | 9 +--
x86_64.c | 230 +++++++++++++++++++++++++++------------------------------------
3 files changed, 123 insertions(+), 171 deletions(-)
diff --git a/defs.h b/defs.h
index 8082d6c..c055e48 100644
--- a/defs.h
+++ b/defs.h
@@ -3330,56 +3330,44 @@ struct arm64_stackframe {
#define VTOP(X) x86_64_VTOP((ulong)(X))
#define IS_VMALLOC_ADDR(X) x86_64_IS_VMALLOC_ADDR((ulong)(X))
-#define PML4_SHIFT 39
-#define PTRS_PER_PML4 512
+/* origin level page */
#define PGDIR_SHIFT 30
#define PTRS_PER_PGD 512
#define PMD_SHIFT 21
#define PTRS_PER_PMD 512
#define PTRS_PER_PTE 512
+/* 4 level page */
+#define PGDIR_SHIFT_4LEVEL 39
+#define PTRS_PER_PGD_4LEVEL 512
+#define PUD_SHIFT 30
+#define PTRS_PER_PUD 512
+
+/* 5 level page */
#define PGDIR_SHIFT_5LEVEL 48
#define PTRS_PER_PGD_5LEVEL 512
#define P4D_SHIFT 39
#define PTRS_PER_P4D 512
#define __PGDIR_SHIFT (machdep->machspec->pgdir_shift)
+#define __PTRS_PER_PGD (machdep->machspec->ptrs_per_pgd)
-#define pml4_index(address) (((address) >> PML4_SHIFT) & (PTRS_PER_PML4-1))
+#define pgd_index(address) (((address) >> __PGDIR_SHIFT) &
(__PTRS_PER_PGD-1))
#define p4d_index(address) (((address) >> P4D_SHIFT) & (PTRS_PER_P4D - 1))
-#define pgd_index(address) (((address) >> __PGDIR_SHIFT) & (PTRS_PER_PGD-1))
+#define pud_index(address) (((address) >> PUD_SHIFT) & (PTRS_PER_PUD - 1))
#define pmd_index(address) (((address) >> PMD_SHIFT) & (PTRS_PER_PMD-1))
#define pte_index(address) (((address) >> PAGE_SHIFT) & (PTRS_PER_PTE - 1))
-#define IS_LAST_PML4_READ(pml4) ((ulong)(pml4) ==
machdep->machspec->last_pml4_read)
-
-#define FILL_PML4() \
- if (!(pc->flags & RUNTIME) || ACTIVE()) { \
- if (!IS_LAST_PML4_READ(vt->kernel_pgd[0])) { \
- readmem(vt->kernel_pgd[0], KVADDR, machdep->machspec->pml4, \
- PAGESIZE(), "init_level4_pgt", FAULT_ON_ERROR); \
- machdep->machspec->last_pml4_read = (ulong)(vt->kernel_pgd[0]); \
- } \
+#define FILL_TOP_PGD() \
+ if (!(pc->flags & RUNTIME) || ACTIVE()) { \
+ FILL_PGD(vt->kernel_pgd[0], KVADDR, PAGESIZE()); \
}
-#define FILL_PML4_HYPER() \
- if (!machdep->machspec->last_pml4_read) { \
- unsigned long idle_pg_table = symbol_exists("idle_pg_table_4") ? \
- symbol_value("idle_pg_table_4") : \
- symbol_value("idle_pg_table"); \
- readmem(idle_pg_table, KVADDR, machdep->machspec->pml4, PAGESIZE(), \
- "idle_pg_table", FAULT_ON_ERROR); \
- machdep->machspec->last_pml4_read = idle_pg_table; \
- }
-
-#define IS_LAST_UPML_READ(pml) ((ulong)(pml) == machdep->machspec->last_upml_read)
-
-#define FILL_UPML(PML, TYPE, SIZE) \
- if (!IS_LAST_UPML_READ(PML)) { \
- readmem((ulonglong)((ulong)(PML)), TYPE, machdep->machspec->upml, \
- SIZE, "pml page", FAULT_ON_ERROR);
\
- machdep->machspec->last_upml_read = (ulong)(PML); \
- }
+#define FILL_TOP_PGD_HYPER() \
+ unsigned long idle_pg_table = symbol_exists("idle_pg_table_4") ? \
+ symbol_value("idle_pg_table_4") : \
+ symbol_value("idle_pg_table"); \
+ FILL_PGD(idle_pg_table, KVADDR, PAGESIZE());
#define IS_LAST_P4D_READ(p4d) ((ulong)(p4d) == machdep->machspec->last_p4d_read)
@@ -5747,10 +5735,6 @@ struct machine_specific {
ulong modules_vaddr;
ulong modules_end;
ulong phys_base;
- char *pml4;
- char *upml;
- ulong last_upml_read;
- ulong last_pml4_read;
char *irqstack;
ulong irq_eframe_link;
struct x86_64_pt_regs_offsets pto;
@@ -5765,6 +5749,7 @@ struct machine_specific {
ulong kernel_image_size;
ulong physical_mask_shift;
ulong pgdir_shift;
+ ulong ptrs_per_pgd;
char *p4d;
ulong last_p4d_read;
struct ORC_data orc;
diff --git a/sadump.c b/sadump.c
index 6b912d4..a189c2d 100644
--- a/sadump.c
+++ b/sadump.c
@@ -2051,10 +2051,11 @@ sadump_calc_kaslr_offset(ulong *kaslr_offset)
* TODO: XEN and 5-level is not supported
*/
vt->kernel_pgd[0] = cr3;
- machdep->machspec->last_pml4_read = vt->kernel_pgd[0];
+ machdep->last_pgd_read = vt->kernel_pgd[0];
machdep->machspec->physical_mask_shift = __PHYSICAL_MASK_SHIFT_2_6;
- machdep->machspec->pgdir_shift = PGDIR_SHIFT;
- if (!readmem(cr3, PHYSADDR, machdep->machspec->pml4, PAGESIZE(),
+ machdep->machspec->pgdir_shift = PGDIR_SHIFT_4LEVEL;
+ machdep->machspec->ptrs_per_pgd = PTRS_PER_PGD_4LEVEL;
+ if (!readmem(cr3, PHYSADDR, machdep->pgd, PAGESIZE(),
"cr3", RETURN_ON_ERROR))
goto quit;
@@ -2099,7 +2100,7 @@ sadump_calc_kaslr_offset(ulong *kaslr_offset)
ret = TRUE;
quit:
vt->kernel_pgd[0] = 0;
- machdep->machspec->last_pml4_read = 0;
+ machdep->last_pgd_read = 0;
return ret;
}
#else
diff --git a/x86_64.c b/x86_64.c
index 7134b4b..0f2e24a 100644
--- a/x86_64.c
+++ b/x86_64.c
@@ -166,20 +166,17 @@ x86_64_init(int when)
machdep->pageoffset = machdep->pagesize - 1;
machdep->pagemask = ~((ulonglong)machdep->pageoffset);
machdep->stacksize = machdep->pagesize * 2;
- if ((machdep->machspec->upml = (char *)malloc(PAGESIZE())) ==
NULL)
- error(FATAL, "cannot malloc upml space.");
- if ((machdep->pgd = (char *)malloc(PAGESIZE())) == NULL)
+ if ((machdep->pgd = (char *)malloc(PAGESIZE())) == NULL)
error(FATAL, "cannot malloc pgd space.");
+ if ((machdep->pud = (char *)malloc(PAGESIZE())) == NULL)
+ error(FATAL, "cannot malloc pud space.");
if ((machdep->pmd = (char *)malloc(PAGESIZE())) == NULL)
error(FATAL, "cannot malloc pmd space.");
if ((machdep->ptbl = (char *)malloc(PAGESIZE())) == NULL)
error(FATAL, "cannot malloc ptbl space.");
- if ((machdep->machspec->pml4 =
- (char *)malloc(PAGESIZE()*2)) == NULL)
- error(FATAL, "cannot malloc pml4 space.");
- machdep->machspec->last_upml_read = 0;
- machdep->machspec->last_pml4_read = 0;
+
machdep->last_pgd_read = 0;
+ machdep->last_pud_read = 0;
machdep->last_pmd_read = 0;
machdep->last_ptbl_read = 0;
machdep->verify_paddr = x86_64_verify_paddr;
@@ -233,12 +230,10 @@ x86_64_init(int when)
machdep->machspec->modules_vaddr = MODULES_VADDR_ORIG;
machdep->machspec->modules_end = MODULES_END_ORIG;
- free(machdep->machspec->upml);
- machdep->machspec->upml = NULL;
-
machdep->uvtop = x86_64_uvtop;
machdep->machspec->physical_mask_shift = __PHYSICAL_MASK_SHIFT_2_6;
machdep->machspec->pgdir_shift = PGDIR_SHIFT;
+ machdep->machspec->ptrs_per_pgd = PTRS_PER_PGD;
break;
case VM_2_6_11:
@@ -264,7 +259,8 @@ x86_64_init(int when)
machdep->uvtop = x86_64_uvtop_level4;
machdep->machspec->physical_mask_shift = __PHYSICAL_MASK_SHIFT_2_6;
- machdep->machspec->pgdir_shift = PGDIR_SHIFT;
+ machdep->machspec->pgdir_shift = PGDIR_SHIFT_4LEVEL;
+ machdep->machspec->ptrs_per_pgd = PTRS_PER_PGD_4LEVEL;
break;
case VM_XEN:
@@ -276,7 +272,6 @@ x86_64_init(int when)
machdep->machspec->modules_vaddr = MODULES_VADDR_XEN;
machdep->machspec->modules_end = MODULES_END_XEN;
machdep->machspec->physical_mask_shift = __PHYSICAL_MASK_SHIFT_XEN;
- machdep->machspec->pgdir_shift = PGDIR_SHIFT;
break;
case VM_XEN_RHEL4:
@@ -288,7 +283,6 @@ x86_64_init(int when)
machdep->machspec->modules_vaddr =
MODULES_VADDR_XEN_RHEL4;
machdep->machspec->modules_end = MODULES_END_XEN_RHEL4;
machdep->machspec->physical_mask_shift = __PHYSICAL_MASK_SHIFT_XEN;
- machdep->machspec->pgdir_shift = PGDIR_SHIFT;
break;
case VM_5LEVEL:
@@ -302,6 +296,7 @@ x86_64_init(int when)
machdep->machspec->vmemmap_end = VMEMMAP_END_5LEVEL;
machdep->machspec->physical_mask_shift = __PHYSICAL_MASK_SHIFT_5LEVEL;
machdep->machspec->pgdir_shift = PGDIR_SHIFT_5LEVEL;
+ machdep->machspec->ptrs_per_pgd = PTRS_PER_PGD_5LEVEL;
if ((machdep->machspec->p4d = (char *)malloc(PAGESIZE())) == NULL)
error(FATAL, "cannot malloc p4d space.");
machdep->machspec->last_p4d_read = 0;
@@ -629,14 +624,21 @@ x86_64_init(int when)
case VM_XEN:
case VM_2_6_11:
machdep->uvtop = x86_64_uvtop_level4_xen_wpt;
+ machdep->machspec->pgdir_shift = PGDIR_SHIFT_4LEVEL;
+ machdep->machspec->ptrs_per_pgd = PTRS_PER_PGD_4LEVEL;
break;
case VM_XEN_RHEL4:
machdep->uvtop = x86_64_uvtop_level4_rhel4_xen_wpt;
+ machdep->machspec->pgdir_shift = PGDIR_SHIFT;
+ machdep->machspec->ptrs_per_pgd = PTRS_PER_PGD;
break;
}
machdep->machspec->physical_mask_shift = __PHYSICAL_MASK_SHIFT_XEN;
- } else
- machdep->uvtop = x86_64_uvtop_level4;
+ } else {
+ machdep->uvtop = x86_64_uvtop_level4;
+ machdep->machspec->pgdir_shift = PGDIR_SHIFT_4LEVEL;
+ machdep->machspec->ptrs_per_pgd = PTRS_PER_PGD_4LEVEL;
+ }
MEMBER_OFFSET_INIT(vcpu_guest_context_user_regs,
"vcpu_guest_context", "user_regs");
ASSIGN_OFFSET(cpu_user_regs_rsp) =
@@ -811,9 +813,11 @@ x86_64_dump_machdep_table(ulong arg)
fprintf(fp, " value_to_symbol: x86_64_value_to_symbol()\n");
fprintf(fp, " in_alternate_stack: x86_64_in_alternate_stack()\n");
fprintf(fp, " last_pgd_read: %lx\n", machdep->last_pgd_read);
+ fprintf(fp, " last_pud_read: %lx\n", machdep->last_pud_read);
fprintf(fp, " last_pmd_read: %lx\n", machdep->last_pmd_read);
fprintf(fp, " last_ptbl_read: %lx\n", machdep->last_ptbl_read);
fprintf(fp, " pgd: %lx\n", (ulong)machdep->pgd);
+ fprintf(fp, " pud: %lx\n", (ulong)machdep->pud);
fprintf(fp, " pmd: %lx\n", (ulong)machdep->pmd);
fprintf(fp, " ptbl: %lx\n", (ulong)machdep->ptbl);
fprintf(fp, " ptrs_per_pgd: %d\n", machdep->ptrs_per_pgd);
@@ -848,15 +852,7 @@ x86_64_dump_machdep_table(ulong arg)
fprintf(fp, " pgdir_shift: %ld\n", ms->pgdir_shift);
fprintf(fp, " GART_start: %lx\n", ms->GART_start);
fprintf(fp, " GART_end: %lx\n", ms->GART_end);
- fprintf(fp, " pml4: %lx\n", (ulong)ms->pml4);
- fprintf(fp, " last_pml4_read: %lx\n",
(ulong)ms->last_pml4_read);
- if (ms->upml) {
- fprintf(fp, " upml: %lx\n", (ulong)ms->upml);
- fprintf(fp, " last_upml_read: %lx\n",
(ulong)ms->last_upml_read);
- } else {
- fprintf(fp, " upml: (unused)\n");
- fprintf(fp, " last_upml_read: (unused)\n");
- }
+
if (ms->p4d) {
fprintf(fp, " p4d: %lx\n", (ulong)ms->p4d);
fprintf(fp, " last_p4d_read: %lx\n",
(ulong)ms->last_p4d_read);
@@ -1511,7 +1507,7 @@ x86_64_init_kernel_pgd(void)
for (i = 0; i < NR_CPUS; i++)
vt->kernel_pgd[i] = kernel_pgt;
- FILL_PML4();
+ FILL_TOP_PGD();
}
/*
@@ -1571,8 +1567,8 @@ x86_64_kpgd_offset(ulong kvaddr, int verbose , int IS_XEN)
{
ulong *pgd;
- FILL_PML4();
- pgd = ((ulong *)machdep->machspec->pml4) + pml4_index(kvaddr);
+ FILL_TOP_PGD();
+ pgd = ((ulong *)machdep->pgd) + pgd_index(kvaddr);
if (verbose) {
fprintf(fp, "PGD DIRECTORY: %lx\n", vt->kernel_pgd[0]);
if(IS_XEN)
@@ -1602,9 +1598,9 @@ x86_64_upgd_offset(struct task_context *tc, ulong uvaddr, int
verbose , int IS_X
sizeof(long), "mm_struct pgd", FAULT_ON_ERROR);
pgd_paddr = x86_64_VTOP((ulong)pgd);
- FILL_UPML(pgd_paddr, PHYSADDR, PAGESIZE());
- pgd = ((ulong *)pgd_paddr) + pml4_index(uvaddr);
- pgd_pte = ULONG(machdep->machspec->pml4 + PAGEOFFSET(pgd));
+ FILL_PGD(pgd_paddr, PHYSADDR, PAGESIZE());
+ pgd = ((ulong *)pgd_paddr) + pgd_index(uvaddr);
+ pgd_pte = ULONG(machdep->pgd + PAGEOFFSET(pgd));
if (verbose) {
if(IS_XEN)
fprintf(fp, " PGD: %lx => %lx [machine]\n", (ulong)pgd, pgd_pte);
@@ -1635,7 +1631,7 @@ x86_64_pud_offset(ulong pgd_pte, ulong vaddr, int verbose , int
IS_XEN)
}
FILL_PUD(pud_paddr, PHYSADDR, PAGESIZE());
- pud = ((ulong *)pud_paddr) + pgd_index(vaddr);
+ pud = ((ulong *)pud_paddr) + pud_index(vaddr);
pud_pte = ULONG(machdep->pud + PAGEOFFSET(pud));
if (verbose) {
if(IS_XEN)
@@ -1917,9 +1913,6 @@ no_upage:
static int
x86_64_uvtop_level4_rhel4_xen_wpt(struct task_context *tc, ulong uvaddr, physaddr_t
*paddr, int verbose)
{
- ulong mm;
- ulong *pgd;
- ulong pgd_paddr;
ulong pgd_pte;
ulong pmd_pte;
ulong pseudo_pmd_pte;
@@ -1936,18 +1929,7 @@ x86_64_uvtop_level4_rhel4_xen_wpt(struct task_context *tc, ulong
uvaddr, physadd
if (IS_KVADDR(uvaddr))
return x86_64_kvtop(tc, uvaddr, paddr, verbose);
- if ((mm = task_mm(tc->task, TRUE)))
- pgd = ULONG_PTR(tt->mm_struct + OFFSET(mm_struct_pgd));
- else
- readmem(tc->mm_struct + OFFSET(mm_struct_pgd), KVADDR, &pgd,
- sizeof(long), "mm_struct pgd", FAULT_ON_ERROR);
-
- pgd_paddr = x86_64_VTOP((ulong)pgd);
- FILL_PGD(pgd_paddr, PHYSADDR, PAGESIZE());
- pgd = ((ulong *)pgd_paddr) + pgd_index(uvaddr);
- pgd_pte = ULONG(machdep->pgd + PAGEOFFSET(pgd));
- if (verbose)
- fprintf(fp, " PGD: %lx => %lx [machine]\n", (ulong)pgd,
pgd_pte);
+ pgd_pte = x86_64_upgd_offset(tc, uvaddr, verbose, TRUE);
if (!(pgd_pte & _PAGE_PRESENT))
goto no_upage;
@@ -2028,9 +2010,6 @@ no_upage:
static int
x86_64_uvtop(struct task_context *tc, ulong uvaddr, physaddr_t *paddr, int verbose)
{
- ulong mm;
- ulong *pgd;
- ulong pgd_paddr;
ulong pgd_pte;
ulong pmd_pte;
ulong pte;
@@ -2047,18 +2026,7 @@ x86_64_uvtop(struct task_context *tc, ulong uvaddr, physaddr_t
*paddr, int verbo
/*
* pgd = pgd_offset(mm, address);
*/
- if ((mm = task_mm(tc->task, TRUE)))
- pgd = ULONG_PTR(tt->mm_struct + OFFSET(mm_struct_pgd));
- else
- readmem(tc->mm_struct + OFFSET(mm_struct_pgd), KVADDR, &pgd,
- sizeof(long), "mm_struct pgd", FAULT_ON_ERROR);
-
- pgd_paddr = x86_64_VTOP((ulong)pgd);
- FILL_PGD(pgd_paddr, PHYSADDR, PAGESIZE());
- pgd = ((ulong *)pgd_paddr) + pgd_index(uvaddr);
- pgd_pte = ULONG(machdep->pgd + PAGEOFFSET(pgd));
- if (verbose)
- fprintf(fp, " PGD: %lx => %lx\n", (ulong)pgd, pgd_pte);
+ pgd_pte = x86_64_upgd_offset(tc, uvaddr, verbose, FALSE);
if (!(pgd_pte & _PAGE_PRESENT))
goto no_upage;
@@ -2119,7 +2087,7 @@ no_upage:
static int
x86_64_kvtop(struct task_context *tc, ulong kvaddr, physaddr_t *paddr, int verbose)
{
- ulong *pml4;
+ ulong *pgd;
ulong pud_pte;
ulong pmd_pte;
ulong pte;
@@ -2148,11 +2116,11 @@ x86_64_kvtop(struct task_context *tc, ulong kvaddr, physaddr_t
*paddr, int verbo
*paddr = kvaddr - DIRECTMAP_VIRT_START;
return TRUE;
}
- FILL_PML4_HYPER();
- pml4 = ((ulong *)machdep->machspec->pml4) + pml4_index(kvaddr);
+ FILL_TOP_PGD_HYPER();
+ pgd = ((ulong *)machdep->pgd) + pgd_index(kvaddr);
if (verbose) {
- fprintf(fp, "PML4 DIRECTORY: %lx\n", vt->kernel_pgd[0]);
- fprintf(fp, "PAGE DIRECTORY: %lx\n", *pml4);
+ fprintf(fp, "PGD DIRECTORY: %lx\n", vt->kernel_pgd[0]);
+ fprintf(fp, "PAGE DIRECTORY: %lx\n", *pgd);
}
} else {
if (!vt->vmalloc_start) {
@@ -2179,10 +2147,10 @@ x86_64_kvtop(struct task_context *tc, ulong kvaddr, physaddr_t
*paddr, int verbo
}
start_vtop_with_pagetable:
- if (!(*pml4 & _PAGE_PRESENT))
+ if (!(*pgd & _PAGE_PRESENT))
goto no_kpage;
- pud_pte = x86_64_pud_offset(*pml4, kvaddr, verbose, FALSE);
+ pud_pte = x86_64_pud_offset(*pgd, kvaddr, verbose, FALSE);
if (!(pud_pte & _PAGE_PRESENT))
goto no_kpage;
@@ -2242,7 +2210,7 @@ x86_64_kvtop_5level(struct task_context *tc, ulong kvaddr,
physaddr_t *paddr, in
static int
x86_64_kvtop_xen_wpt(struct task_context *tc, ulong kvaddr, physaddr_t *paddr, int
verbose)
{
- ulong *pml4;
+ ulong *pgd;
ulong pud_pte;
ulong pmd_pte;
ulong pseudo_pmd_pte;
@@ -2254,11 +2222,11 @@ x86_64_kvtop_xen_wpt(struct task_context *tc, ulong kvaddr,
physaddr_t *paddr, i
/*
* pgd = pgd_offset_k(addr);
*/
- pml4 = x86_64_kpgd_offset(kvaddr, verbose, TRUE);
- if (!(*pml4 & _PAGE_PRESENT))
+ pgd = x86_64_kpgd_offset(kvaddr, verbose, TRUE);
+ if (!(*pgd & _PAGE_PRESENT))
goto no_kpage;
- pud_pte = x86_64_pud_offset(*pml4, kvaddr, verbose, TRUE);
+ pud_pte = x86_64_pud_offset(*pgd, kvaddr, verbose, TRUE);
if (!(pud_pte & _PAGE_PRESENT))
goto no_kpage;
@@ -5827,7 +5795,8 @@ parse_cmdline_args(void)
void
x86_64_clear_machdep_cache(void)
{
- machdep->machspec->last_upml_read = 0;
+ if (machdep->last_pgd_read != vt->kernel_pgd[0])
+ machdep->last_pgd_read = 0;
}
#define PUSH_RBP_MOV_RSP_RBP 0xe5894855
@@ -6206,12 +6175,12 @@ use_cr3:
if (CRASHDEBUG(1))
fprintf(fp, "x86_64_xen_kdump_p2m_create: cr3: %lx\n",
xkd->cr3);
- if (!readmem(PTOB(xkd->cr3), PHYSADDR, machdep->machspec->pml4,
+ if (!readmem(PTOB(xkd->cr3), PHYSADDR, machdep->pgd,
PAGESIZE(), "xen kdump cr3 page", RETURN_ON_ERROR))
error(FATAL, "cannot read xen kdump cr3 page\n");
if (CRASHDEBUG(7))
- x86_64_debug_dump_page(fp, machdep->machspec->pml4,
+ x86_64_debug_dump_page(fp, machdep->pgd,
"contents of PML4 page:");
/*
@@ -6246,7 +6215,7 @@ use_cr3:
if (CRASHDEBUG(1))
fprintf(fp, "phys_to_machine_mapping: %lx\n", kvaddr);
- machdep->last_pgd_read = BADADDR;
+ machdep->last_pud_read = BADADDR;
machdep->last_pmd_read = BADADDR;
machdep->last_ptbl_read = BADADDR;
@@ -6261,7 +6230,7 @@ use_cr3:
fprintf(fp, "\n");
}
- machdep->last_pgd_read = 0;
+ machdep->last_pud_read = 0;
machdep->last_ptbl_read = 0;
machdep->last_pmd_read = 0;
pc->curcmd_flags &= ~XEN_MACHINE_ADDR;
@@ -6275,33 +6244,33 @@ static char *
x86_64_xen_kdump_load_page(ulong kvaddr, char *pgbuf)
{
ulong mfn;
- ulong *pml4, *pgd, *pmd, *ptep;
+ ulong *pgd, *pud, *pmd, *ptep;
- pml4 = ((ulong *)machdep->machspec->pml4) + pml4_index(kvaddr);
- mfn = ((*pml4) & PHYSICAL_PAGE_MASK) >> PAGESHIFT();
+ pgd = ((ulong *)machdep->pgd) + pgd_index(kvaddr);
+ mfn = ((*pgd) & PHYSICAL_PAGE_MASK) >> PAGESHIFT();
if (CRASHDEBUG(3))
fprintf(fp,
- "[%lx] pml4: %lx mfn: %lx pml4_index: %lx\n",
- kvaddr, *pml4, mfn, pml4_index(kvaddr));
+ "[%lx] pgd: %lx mfn: %lx pgd_index: %lx\n",
+ kvaddr, *pgd, mfn, pgd_index(kvaddr));
- if (!readmem(PTOB(mfn), PHYSADDR, machdep->pgd, PAGESIZE(),
+ if (!readmem(PTOB(mfn), PHYSADDR, machdep->pud, PAGESIZE(),
"xen kdump pud page", RETURN_ON_ERROR))
error(FATAL, "cannot read/find pud page\n");
- machdep->last_pgd_read = mfn;
+ machdep->last_pud_read = mfn;
if (CRASHDEBUG(7))
- x86_64_debug_dump_page(fp, machdep->pgd,
+ x86_64_debug_dump_page(fp, machdep->pud,
"contents of page upper directory page:");
- pgd = ((ulong *)machdep->pgd) + pgd_index(kvaddr);
- mfn = ((*pgd) & PHYSICAL_PAGE_MASK) >> PAGESHIFT();
+ pud = ((ulong *)machdep->pud) + pud_index(kvaddr);
+ mfn = ((*pud) & PHYSICAL_PAGE_MASK) >> PAGESHIFT();
if (CRASHDEBUG(3))
fprintf(fp,
- "[%lx] pgd: %lx mfn: %lx pgd_index: %lx\n",
- kvaddr, *pgd, mfn, pgd_index(kvaddr));
+ "[%lx] pud: %lx mfn: %lx pud_index: %lx\n",
+ kvaddr, *pgd, mfn, pud_index(kvaddr));
if (!readmem(PTOB(mfn), PHYSADDR, machdep->pmd, PAGESIZE(),
"xen kdump pmd page", RETURN_ON_ERROR))
@@ -6354,21 +6323,21 @@ static ulong
x86_64_xen_kdump_page_mfn(ulong kvaddr)
{
ulong mfn;
- ulong *pml4, *pgd, *pmd, *ptep;
+ ulong *pgd, *pud, *pmd, *ptep;
- pml4 = ((ulong *)machdep->machspec->pml4) + pml4_index(kvaddr);
- mfn = ((*pml4) & PHYSICAL_PAGE_MASK) >> PAGESHIFT();
+ pgd = ((ulong *)machdep->pgd) + pgd_index(kvaddr);
+ mfn = ((*pgd) & PHYSICAL_PAGE_MASK) >> PAGESHIFT();
- if ((mfn != machdep->last_pgd_read) &&
- !readmem(PTOB(mfn), PHYSADDR, machdep->pgd, PAGESIZE(),
+ if ((mfn != machdep->last_pud_read) &&
+ !readmem(PTOB(mfn), PHYSADDR, machdep->pud, PAGESIZE(),
"xen kdump pud entry", RETURN_ON_ERROR))
error(FATAL, "cannot read/find pud page\n");
- machdep->last_pgd_read = mfn;
+ machdep->last_pud_read = mfn;
- pgd = ((ulong *)machdep->pgd) + pgd_index(kvaddr);
- mfn = ((*pgd) & PHYSICAL_PAGE_MASK) >> PAGESHIFT();
+ pud = ((ulong *)machdep->pud) + pud_index(kvaddr);
+ mfn = ((*pud) & PHYSICAL_PAGE_MASK) >> PAGESHIFT();
- if ((mfn != machdep->last_pmd_read) &&
+ if ((mfn != machdep->last_pmd_read) &&
!readmem(PTOB(mfn), PHYSADDR, machdep->pmd, PAGESIZE(),
"xen kdump pmd entry", RETURN_ON_ERROR))
error(FATAL, "cannot read/find pmd page\n");
@@ -6723,12 +6692,12 @@ x86_64_xendump_p2m_create(struct xendump_data *xd)
mfn = ctrlreg[3] >> PAGESHIFT();
- if (!xc_core_mfn_to_page(mfn, machdep->machspec->pml4))
+ if (!xc_core_mfn_to_page(mfn, machdep->pgd))
error(FATAL, "cannot read/find cr3 page\n");
if (CRASHDEBUG(7))
- x86_64_debug_dump_page(xd->ofp, machdep->machspec->pml4,
- "contents of PML4 page:");
+ x86_64_debug_dump_page(xd->ofp, machdep->pgd,
+ "contents of PGD page:");
/*
* kernel version < 2.6.27 => end_pfn
@@ -6809,12 +6778,12 @@ x86_64_pvops_xendump_p2m_create(struct xendump_data *xd)
mfn = ctrlreg[3] >> PAGESHIFT();
- if (!xc_core_mfn_to_page(mfn, machdep->machspec->pml4))
+ if (!xc_core_mfn_to_page(mfn, machdep->pgd))
error(FATAL, "cannot read/find cr3 page\n");
if (CRASHDEBUG(7))
- x86_64_debug_dump_page(xd->ofp, machdep->machspec->pml4,
- "contents of PML4 page:");
+ x86_64_debug_dump_page(xd->ofp, machdep->pgd,
+ "contents of PGD page:");
/*
* kernel version < 2.6.27 => end_pfn
@@ -6990,32 +6959,32 @@ static char *
x86_64_xendump_load_page(ulong kvaddr, struct xendump_data *xd)
{
ulong mfn;
- ulong *pml4, *pgd, *pmd, *ptep;
+ ulong *pgd, *pud, *pmd, *ptep;
- pml4 = ((ulong *)machdep->machspec->pml4) + pml4_index(kvaddr);
- mfn = ((*pml4) & PHYSICAL_PAGE_MASK) >> PAGESHIFT();
+ pgd = ((ulong *)machdep->pgd) + pgd_index(kvaddr);
+ mfn = ((*pgd) & PHYSICAL_PAGE_MASK) >> PAGESHIFT();
if (CRASHDEBUG(3))
fprintf(xd->ofp,
- "[%lx] pml4: %lx mfn: %lx pml4_index: %lx\n",
- kvaddr, *pml4, mfn, pml4_index(kvaddr));
+ "[%lx] pgd: %lx mfn: %lx pgd_index: %lx\n",
+ kvaddr, *pgd, mfn, pgd_index(kvaddr));
- if (!xc_core_mfn_to_page(mfn, machdep->pgd))
+ if (!xc_core_mfn_to_page(mfn, machdep->pud))
error(FATAL, "cannot read/find pud page\n");
- machdep->last_pgd_read = mfn;
+ machdep->last_pud_read = mfn;
if (CRASHDEBUG(7))
- x86_64_debug_dump_page(xd->ofp, machdep->pgd,
+ x86_64_debug_dump_page(xd->ofp, machdep->pud,
"contents of page upper directory page:");
- pgd = ((ulong *)machdep->pgd) + pgd_index(kvaddr);
- mfn = ((*pgd) & PHYSICAL_PAGE_MASK) >> PAGESHIFT();
+ pud = ((ulong *)machdep->pud) + pud_index(kvaddr);
+ mfn = ((*pud) & PHYSICAL_PAGE_MASK) >> PAGESHIFT();
if (CRASHDEBUG(3))
fprintf(xd->ofp,
- "[%lx] pgd: %lx mfn: %lx pgd_index: %lx\n",
- kvaddr, *pgd, mfn, pgd_index(kvaddr));
+ "[%lx] pud: %lx mfn: %lx pud_index: %lx\n",
+ kvaddr, *pud, mfn, pud_index(kvaddr));
if (!xc_core_mfn_to_page(mfn, machdep->pmd))
error(FATAL, "cannot read/find pmd page\n");
@@ -7069,18 +7038,18 @@ x86_64_xendump_page_index(ulong kvaddr, struct xendump_data *xd)
{
int idx;
ulong mfn;
- ulong *pml4, *pgd, *pmd, *ptep;
+ ulong *pgd, *pud, *pmd, *ptep;
- pml4 = ((ulong *)machdep->machspec->pml4) + pml4_index(kvaddr);
- mfn = ((*pml4) & PHYSICAL_PAGE_MASK) >> PAGESHIFT();
+ pgd = ((ulong *)machdep->pgd) + pgd_index(kvaddr);
+ mfn = ((*pgd) & PHYSICAL_PAGE_MASK) >> PAGESHIFT();
- if ((mfn != machdep->last_pgd_read) &&
- !xc_core_mfn_to_page(mfn, machdep->pgd))
+ if ((mfn != machdep->last_pud_read) &&
+ !xc_core_mfn_to_page(mfn, machdep->pud))
error(FATAL, "cannot read/find pud page\n");
- machdep->last_pgd_read = mfn;
+ machdep->last_pud_read = mfn;
- pgd = ((ulong *)machdep->pgd) + pgd_index(kvaddr);
- mfn = ((*pgd) & PHYSICAL_PAGE_MASK) >> PAGESHIFT();
+ pud = ((ulong *)machdep->pud) + pud_index(kvaddr);
+ mfn = ((*pud) & PHYSICAL_PAGE_MASK) >> PAGESHIFT();
if ((mfn != machdep->last_pmd_read) &&
!xc_core_mfn_to_page(mfn, machdep->pmd))
@@ -7571,20 +7540,17 @@ x86_64_init_hyper(int when)
machdep->pageoffset = machdep->pagesize - 1;
machdep->pagemask = ~((ulonglong)machdep->pageoffset);
machdep->stacksize = machdep->pagesize * 2;
- if ((machdep->machspec->upml = (char *)malloc(PAGESIZE())) ==
NULL)
- error(FATAL, "cannot malloc upml space.");
if ((machdep->pgd = (char *)malloc(PAGESIZE())) == NULL)
error(FATAL, "cannot malloc pgd space.");
+ if ((machdep->pud = (char *)malloc(PAGESIZE())) == NULL)
+ error(FATAL, "cannot malloc pud space.");
if ((machdep->pmd = (char *)malloc(PAGESIZE())) == NULL)
error(FATAL, "cannot malloc pmd space.");
if ((machdep->ptbl = (char *)malloc(PAGESIZE())) == NULL)
error(FATAL, "cannot malloc ptbl space.");
- if ((machdep->machspec->pml4 =
- (char *)malloc(PAGESIZE()*2)) == NULL)
- error(FATAL, "cannot malloc pml4 space.");
- machdep->machspec->last_upml_read = 0;
- machdep->machspec->last_pml4_read = 0;
+
machdep->last_pgd_read = 0;
+ machdep->last_pud_read = 0;
machdep->last_pmd_read = 0;
machdep->last_ptbl_read = 0;
machdep->verify_paddr = generic_verify_paddr;
--
2.14.3
1:31 a.m.
New subject: [PATCH 4/5] x86_64.c: Add the VMEMMAP support for 5 level page table
Append the "VMEMMAP" flag to machdep's flags too.
Signed-off-by: Dou Liyang <douly.fnst(a)cn.fujitsu.com>
---
x86_64.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/x86_64.c b/x86_64.c
index 0f2e24a..8914830 100644
--- a/x86_64.c
+++ b/x86_64.c
@@ -294,6 +294,8 @@ x86_64_init(int when)
machdep->machspec->modules_end = MODULES_END_5LEVEL;
machdep->machspec->vmemmap_vaddr = VMEMMAP_VADDR_5LEVEL;
machdep->machspec->vmemmap_end = VMEMMAP_END_5LEVEL;
+ if (symbol_exists("vmemmap_populate"))
+ machdep->flags |= VMEMMAP;
machdep->machspec->physical_mask_shift = __PHYSICAL_MASK_SHIFT_5LEVEL;
machdep->machspec->pgdir_shift = PGDIR_SHIFT_5LEVEL;
machdep->machspec->ptrs_per_pgd = PTRS_PER_PGD_5LEVEL;
--
2.14.3
1:31 a.m.
New subject: [PATCH 5/5] x86_64: Add 5-level paging support for x86_64_k/uvtop()
Add the support for x86_64 5-level page tables. Squeeze the 5-level support
into the currently existing x86_64_kvtop() and x86_64_uvtop() functions instead
of using the new xxx_5level() function placeholders.
Signed-off-by: Dou Liyang <douly.fnst(a)cn.fujitsu.com>
---
x86_64.c | 90 +++++++++++++++++++++++++++++++++++++++++++++-------------------
1 file changed, 63 insertions(+), 27 deletions(-)
diff --git a/x86_64.c b/x86_64.c
index 8914830..944c7ee 100644
--- a/x86_64.c
+++ b/x86_64.c
@@ -19,13 +19,11 @@
#ifdef X86_64
static int x86_64_kvtop(struct task_context *, ulong, physaddr_t *, int);
-static int x86_64_kvtop_5level(struct task_context *, ulong, physaddr_t *, int);
static int x86_64_kvtop_xen_wpt(struct task_context *, ulong, physaddr_t *, int);
static int x86_64_uvtop(struct task_context *, ulong, physaddr_t *, int);
static int x86_64_uvtop_level4(struct task_context *, ulong, physaddr_t *, int);
static int x86_64_uvtop_level4_xen_wpt(struct task_context *, ulong, physaddr_t *, int);
static int x86_64_uvtop_level4_rhel4_xen_wpt(struct task_context *, ulong, physaddr_t *,
int);
-static int x86_64_uvtop_5level(struct task_context *, ulong, physaddr_t *, int);
static int x86_64_task_uses_5level(struct task_context *);
static ulong x86_64_vmalloc_start(void);
static int x86_64_is_task_addr(ulong);
@@ -80,6 +78,7 @@ static int x86_64_is_kvaddr(ulong);
static int x86_64_is_uvaddr(ulong, struct task_context *);
static ulong *x86_64_kpgd_offset(ulong, int, int);
static ulong x86_64_upgd_offset(struct task_context *, ulong, int, int);
+static ulong x86_64_p4d_offset(ulong, ulong, int, int);
static ulong x86_64_pud_offset(ulong, ulong, int, int);
static ulong x86_64_pmd_offset(ulong, ulong, int, int);
static ulong x86_64_pte_offset(ulong, ulong, int, int);
@@ -1613,6 +1612,31 @@ x86_64_upgd_offset(struct task_context *tc, ulong uvaddr, int
verbose , int IS_X
return pgd_pte;
}
+/*
+ * Find an entry in the fourth-level page table..
+ * p4d = p4d_offset(pgd, address);
+ */
+static ulong
+x86_64_p4d_offset(ulong pgd_pte, ulong vaddr, int verbose ,int IS_XEN)
+{
+ ulong *p4d;
+ ulong p4d_paddr;
+ ulong p4d_pte;
+
+ p4d_paddr = pgd_pte & PHYSICAL_PAGE_MASK;
+ FILL_P4D(p4d_paddr, PHYSADDR, PAGESIZE());
+ p4d = ((ulong *)p4d_paddr) + p4d_index(vaddr);
+ p4d_pte = ULONG(machdep->machspec->p4d + PAGEOFFSET(p4d));
+ if (verbose) {
+ if(IS_XEN)
+ fprintf(fp, " P4D: %lx => %lx [machine]\n", (ulong)p4d, p4d_pte);
+ else
+ fprintf(fp, " P4D: %lx => %lx\n", (ulong)p4d, p4d_pte);
+ }
+
+ return p4d_pte;
+}
+
/*
* Find an entry in the third-level page table..
* pud = pud_offset(pgd, address);
@@ -1734,17 +1758,30 @@ x86_64_uvtop_level4(struct task_context *tc, ulong uvaddr,
physaddr_t *paddr, in
if (IS_KVADDR(uvaddr))
return x86_64_kvtop(tc, uvaddr, paddr, verbose);
- if ((machdep->flags & VM_5LEVEL) && x86_64_task_uses_5level(tc))
- return x86_64_uvtop_5level(tc, uvaddr, paddr, verbose);
-
pgd_pte = x86_64_upgd_offset(tc, uvaddr, verbose, FALSE);
if (!(pgd_pte & _PAGE_PRESENT))
goto no_upage;
- /*
- * pud = pud_offset(pgd, address);
- */
- pud_pte = x86_64_pud_offset(pgd_pte, uvaddr, verbose, FALSE);
+ /* If the VM is in 5-level page table*/
+ if (machdep->flags & VM_5LEVEL && x86_64_task_uses_5level(tc)) {
+ ulong p4d_pte;
+ /*
+ * p4d = p4d_offset(pgd, address);
+ */
+ p4d_pte = x86_64_p4d_offset(pgd_pte, uvaddr, verbose, FALSE);
+ if (!(p4d_pte & _PAGE_PRESENT))
+ goto no_upage;
+ /*
+ * pud = pud_offset(p4d, address);
+ */
+ pud_pte = x86_64_pud_offset(p4d_pte, uvaddr, verbose, FALSE);
+ } else {
+ /*
+ * pud = pud_offset(pgd, address);
+ */
+ pud_pte = x86_64_pud_offset(pgd_pte, uvaddr, verbose, FALSE);
+ }
+
if (!(pud_pte & _PAGE_PRESENT))
goto no_upage;
@@ -1803,13 +1840,6 @@ x86_64_task_uses_5level(struct task_context *tc)
return FALSE;
}
-static int
-x86_64_uvtop_5level(struct task_context *tc, ulong uvaddr, physaddr_t *paddr, int
verbose)
-{
- error(INFO, "support for 5-level page tables TBD\n");
- return FALSE;
-}
-
static int
x86_64_uvtop_level4_xen_wpt(struct task_context *tc, ulong uvaddr, physaddr_t *paddr, int
verbose)
{
@@ -2139,9 +2169,6 @@ x86_64_kvtop(struct task_context *tc, ulong kvaddr, physaddr_t
*paddr, int verbo
if (XEN() && (kt->xen_flags & WRITABLE_PAGE_TABLES))
return (x86_64_kvtop_xen_wpt(tc, kvaddr, paddr, verbose));
- if (machdep->flags & VM_5LEVEL)
- return (x86_64_kvtop_5level(tc, kvaddr, paddr, verbose));
-
/*
* pgd = pgd_offset_k(addr);
*/
@@ -2152,7 +2179,23 @@ start_vtop_with_pagetable:
if (!(*pgd & _PAGE_PRESENT))
goto no_kpage;
- pud_pte = x86_64_pud_offset(*pgd, kvaddr, verbose, FALSE);
+ /* If the VM is in 5-level page table*/
+ if (machdep->flags & VM_5LEVEL) {
+ ulong p4d_pte;
+ /*
+ * p4d = p4d_offset(pgd, address);
+ */
+ p4d_pte = x86_64_p4d_offset(*pgd, kvaddr, verbose, FALSE);
+ if (!(p4d_pte & _PAGE_PRESENT))
+ goto no_kpage;
+ /*
+ * pud = pud_offset(p4d, address);
+ */
+ pud_pte = x86_64_pud_offset(p4d_pte, kvaddr, verbose, FALSE);
+ } else {
+ pud_pte = x86_64_pud_offset(*pgd, kvaddr, verbose, FALSE);
+ }
+
if (!(pud_pte & _PAGE_PRESENT))
goto no_kpage;
@@ -2202,13 +2245,6 @@ no_kpage:
return FALSE;
}
-static int
-x86_64_kvtop_5level(struct task_context *tc, ulong kvaddr, physaddr_t *paddr, int
verbose)
-{
- error(INFO, "support for 5-level page tables TBD\n");
- return FALSE;
-}
-
static int
x86_64_kvtop_xen_wpt(struct task_context *tc, ulong kvaddr, physaddr_t *paddr, int
verbose)
{
--
2.14.3
2:56 p.m.
----- Original Message -----
I found Dave had alread done the first phase of future support for
x86_64
5-level page tables(commit 307e7f35f510). when I asked him about the
state of this work, he gave me a more detailed answer and suggestion.
I follow his advice, and do the following job.
1. Refine the original logical:
1) Create some new common function for getting the offset of page table
2) Repace the PML4 and UPML with the common PGD:
machdep->machspec->pml4/upml ==> machdep->pdg
3) Using the PUD in x86_64
2. Add 5-level page tables support for x86_64_k/uvtop()
This patchset is the second phase of the work, As Dave said, we need to be
a manner of determining very early on whether the kernel page tables are
using 5-level and whether each user-space task is using 4- or 5-level page
tables. These will be done after this phase.
About test work:
I have tested this patchset with 4-level and 5-level paging table.
sadump/ Xen/ Old Linux / RHEL4 are not be tested.
Hello Dou,
Thank you very much for the work you have done so far. I have not spent
any time looking at the patches in detail, but instead I first ran a quick
test of the patch on a set of ~250 kernels that I keep around for testing,
where I just ran the "mod" command to at least verify that kernel virtual
addresses could be translated.
Now, as always, backwards compatibility must be maintained. I do not have
any sadump dumpfiles, but obviously you (Fujitsu) can test those. However
I do have some older Xen and RHEL4-era kernels in my sample set.
As it turns out, *all* RHEL4 kernels failed (i.e. any kernel version
earlier than 2.6.9), which report "WARNING: cannot access vmalloc'd
module memory" during initialization when trying to gather the kernel
module list.
For all of the 2.6.9 and earlier kernels, they show the "WARNING: cannot
access vmalloc'd module memory" message during session initialization:
$ crash vmlinux-2.6.9-42.0.2.ELsmp.gz vmcore
crash 7.2.1rc26
Copyright (C) 2002-2017 Red Hat, Inc.
Copyright (C) 2004, 2005, 2006, 2010 IBM Corporation
Copyright (C) 1999-2006 Hewlett-Packard Co
Copyright (C) 2005, 2006, 2011, 2012 Fujitsu Limited
Copyright (C) 2006, 2007 VA Linux Systems Japan K.K.
Copyright (C) 2005, 2011 NEC Corporation
Copyright (C) 1999, 2002, 2007 Silicon Graphics, Inc.
Copyright (C) 1999, 2000, 2001, 2002 Mission Critical Linux, Inc.
This program is free software, covered by the GNU General Public License,
and you are welcome to change it and/or distribute copies of it under
certain conditions. Enter "help copying" to see the conditions.
This program has absolutely no warranty. Enter "help warranty" for details.
GNU gdb (GDB) 7.6
Copyright (C) 2013 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law. Type "show copying"
and "show warranty" for details.
This GDB was configured as "x86_64-unknown-linux-gnu"...
please wait... (gathering module symbol data)
WARNING: cannot access vmalloc'd module memory
KERNEL: vmlinux-2.6.9-42.0.2.ELsmp.gz
DUMPFILE: vmcore
CPUS: 8
DATE: Tue Nov 21 19:14:17 2006
UPTIME: 6 days, 01:23:25
LOAD AVERAGE: 24.34, 7.89, 4.46
TASKS: 865
NODENAME: lonrs00268
RELEASE: 2.6.9-42.0.2.ELsmp
VERSION: #1 SMP Thu Aug 17 17:57:31 EDT 2006
MACHINE: x86_64 (2199 Mhz)
MEMORY: 16 GB
PANIC: "Kernel BUG at panic:75"
PID: 20046
COMMAND: "oracle"
TASK: 101c6b047f0 [THREAD_INFO: 101a428a000]
CPU: 7
STATE: TASK_RUNNING (NMI)
crash>
If I run the session with "crash -d4 vmlinux-2.6.9-42.0.2.ELsmp.gz vmcore",
you can see that it it reads a "pud page", but then fails:
...
please wait... (gathering module symbol data)module: ffffffffa0634180
<readmem: ffffffffa0634180, KVADDR, "module struct", 1408, (ROE|Q),
f73780>
<readmem: 4f8000, PHYSADDR, "pud page", 4096, (FOE), 2080b40>
<read_diskdump: addr: 4f8000 paddr: 4f8000 cnt: 4096>
crash: invalid kernel virtual address: ffffffffa0634180 type: "module
struct"
WARNING: cannot access vmalloc'd module memory
...
Without the patch, the module virtual address translation succeeds:
...
please wait... (gathering module symbol data)module: ffffffffa0634180
<readmem: ffffffffa0634180, KVADDR, "module struct", 1408, (ROE|Q),
f705e0>
<readmem: 103000, PHYSADDR, "pgd page", 4096, (FOE), 25d7b50>
<read_diskdump: addr: 103000 paddr: 103000 cnt: 4096>
<readmem: 105000, PHYSADDR, "pmd page", 4096, (FOE), 25d8b60>
<read_diskdump: addr: 105000 paddr: 105000 cnt: 4096>
<readmem: d9bfb0000, PHYSADDR, "page table", 4096, (FOE), 25d9b70>
<read_diskdump: addr: d9bfb0000 paddr: d9bfb0000 cnt: 4096>
<read_diskdump: addr: ffffffffa0634180 paddr: d9bfb3180 cnt: 1408>
...
So it appears to be reading from the wrong starting page table location,
i.e., from "pud page 4f8000" instead of "pgd page 103000".
Also, several Xen kernels failed with segmentation violations during
session initialization. They all fail here in x86_64_xendump_load_page(),
when "*pgd" gets referenced:
static char *
x86_64_xendump_load_page(ulong kvaddr, struct xendump_data *xd)
{
ulong mfn;
ulong *pgd, *pud, *pmd, *ptep;
pgd = ((ulong *)machdep->pgd) + pgd_index(kvaddr);
mfn = ((*pgd) & PHYSICAL_PAGE_MASK) >> PAGESHIFT();
^^^^
Here is the relevant part of the gdb trace of a 2.6.18-based xen
kernel:
Program terminated with signal 11, Segmentation fault.
#0 0x0000000000502748 in x86_64_xendump_load_page
(kvaddr=kvaddr@entry=18446744071568498888, xd=0xf521a0 <xendump_data>,
xd=0xf521a0 <xendump_data>) at x86_64.c:7003
7003 mfn = ((*pgd) & PHYSICAL_PAGE_MASK) >> PAGESHIFT();
Missing separate debuginfos, use: debuginfo-install glibc-2.17-157.el7.x86_64
libgcc-4.8.5-11.el7.x86_64 libstdc++-4.8.5-11.el7.x86_64 lzo-2.06-8.el7.x86_64
ncurses-libs-5.9-13.20130511.el7.x86_64 snappy-1.1.0-3.el7.x86_64
xz-libs-5.2.2-1.el7.x86_64 zlib-1.2.7-17.el7.x86_64
(gdb) bt
#0 0x0000000000502748 in x86_64_xendump_load_page
(kvaddr=kvaddr@entry=18446744071568498888, xd=0xf521a0 <xendump_data>,
xd=0xf521a0 <xendump_data>) at x86_64.c:7003
#1 0x0000000000503191 in x86_64_xendump_p2m_create (xd=0xf521a0 <xendump_data>) at
x86_64.c:6749
#2 0x0000000000565d4e in xc_core_create_pfn_tables () at xendump.c:1258
#3 xc_core_read (addr=<optimized out>, paddr=7080864, cnt=32, bufptr=0xf70f80
<shared_bufs>) at xendump.c:168
#4 read_xendump (fd=<optimized out>, bufptr=0xf70f80 <shared_bufs>, cnt=32,
addr=<optimized out>, paddr=7080864) at xendump.c:836
#5 0x000000000047b038 in readmem (addr=18446744071569148832, memtype=memtype@entry=1,
buffer=buffer@entry=0xf70f80 <shared_bufs>,
size=size@entry=32, type=type@entry=0x94dcc3 "possible",
error_handle=error_handle@entry=2) at memory.c:2233
#6 0x00000000004ea33e in cpu_maps_init () at kernel.c:903
#7 kernel_init () at kernel.c:118
#8 0x0000000000467e5a in main_loop () at main.c:768
#9 0x000000000069dad3 in captured_command_loop (data=data@entry=0x0) at main.c:258
#10 0x000000000069c37a in catch_errors (func=func@entry=0x69dac0
<captured_command_loop>, func_args=func_args@entry=0x0,
errstring=errstring@entry=0x8e713f "", mask=mask@entry=6) at
exceptions.c:557
#11 0x000000000069ea66 in captured_main (data=data@entry=0x7ffd637c92a0) at main.c:1064
#12 0x000000000069c37a in catch_errors (func=func@entry=0x69dda0 <captured_main>,
func_args=func_args@entry=0x7ffd637c92a0,
errstring=errstring@entry=0x8e713f "", mask=mask@entry=6) at
exceptions.c:557
#13 0x000000000069edc7 in gdb_main (args=0x7ffd637c92a0) at main.c:1079
#14 gdb_main_entry (argc=<optimized out>, argv=argv@entry=0x7ffd637c9408) at
main.c:1099
#15 0x00000000004f0604 in gdb_main_loop (argc=<optimized out>, argc@entry=3,
argv=argv@entry=0x7ffd637c9408) at gdb_interface.c:76
#16 0x00000000004662c5 in main (argc=3, argv=0x7ffd637c9408) at main.c:707
(gdb) p pgd
$1 = (ulong *) 0xfffffffc054f4210
(gdb)
I haven't investigated further, but in all of the xen cases, the
value of "pgd" above was a kernel virtual address as shown in the
example above.
However, without the patch, the function looks like this, and with
my debug printf of "pml4", the address is a user-space address as
expected:
static char *
x86_64_xendump_load_page(ulong kvaddr, struct xendump_data *xd)
{
ulong mfn;
ulong *pml4, *pgd, *pmd, *ptep;
pml4 = ((ulong *)machdep->machspec->pml4) + pml4_index(kvaddr);
mfn = ((*pml4) & PHYSICAL_PAGE_MASK) >> PAGESHIFT();
fprintf(fp, "x86_64_xendump_load_page: pml4: %lx\n", pml4);
...
So for example, with the debug statement, I see this:
# crash vmlinux-2.6.18-1.2714.el5xen.gz xguest-crashdump
crash 7.2.1rc26
Copyright (C) 2002-2017 Red Hat, Inc.
Copyright (C) 2004, 2005, 2006, 2010 IBM Corporation
Copyright (C) 1999-2006 Hewlett-Packard Co
Copyright (C) 2005, 2006, 2011, 2012 Fujitsu Limited
Copyright (C) 2006, 2007 VA Linux Systems Japan K.K.
Copyright (C) 2005, 2011 NEC Corporation
Copyright (C) 1999, 2002, 2007 Silicon Graphics, Inc.
Copyright (C) 1999, 2000, 2001, 2002 Mission Critical Linux, Inc.
This program is free software, covered by the GNU General Public License,
and you are welcome to change it and/or distribute copies of it under
certain conditions. Enter "help copying" to see the conditions.
This program has absolutely no warranty. Enter "help warranty" for details.
GNU gdb (GDB) 7.6
Copyright (C) 2013 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law. Type "show copying"
and "show warranty" for details.
This GDB was configured as "x86_64-unknown-linux-gnu"...
x86_64_xendump_load_page: pml4: 25d6c08
x86_64_xendump_load_page: pml4: 25d6c08
KERNEL: vmlinux-2.6.18-1.2714.el5xen.gz
DUMPFILE: xguest-crashdump
...
In a private email, I will send you a pointer to where I have temporarily
stored the 2 vmlinux/vmcore pairs shown above. I'm thinking that it will
probably be fairly easy for you to figure out what's happening in both cases.
Again, I very much appreciate the work you have undertaken here.
Thanks,
Dave
7:49 p.m.
Hi Dave,
[...]
Thank you very much for the work you have done so far. I have not
spent
any time looking at the patches in detail, but instead I first ran a quick
test of the patch on a set of ~250 kernels that I keep around for testing,
where I just ran the "mod" command to at least verify that kernel virtual
addresses could be translated.
Now, as always, backwards compatibility must be maintained. I do not have
any sadump dumpfiles, but obviously you (Fujitsu) can test those. However
Yes, I am waiting the machine which can support sadump. I will test the
sadump dumpfiles.
I do have some older Xen and RHEL4-era kernels in my sample set.
Thank you so much about that, I will keep the backwards compatibility.
As it turns out, *all* RHEL4 kernels failed (i.e. any kernel version
earlier than 2.6.9), which report "WARNING: cannot access vmalloc'd
module memory" during initialization when trying to gather the kernel
module list.
For all of the 2.6.9 and earlier kernels, they show the "WARNING: cannot
access vmalloc'd module memory" message during session initialization:
$ crash vmlinux-2.6.9-42.0.2.ELsmp.gz vmcore
crash 7.2.1rc26
Copyright (C) 2002-2017 Red Hat, Inc.
Copyright (C) 2004, 2005, 2006, 2010 IBM Corporation
Copyright (C) 1999-2006 Hewlett-Packard Co
Copyright (C) 2005, 2006, 2011, 2012 Fujitsu Limited
Copyright (C) 2006, 2007 VA Linux Systems Japan K.K.
Copyright (C) 2005, 2011 NEC Corporation
Copyright (C) 1999, 2002, 2007 Silicon Graphics, Inc.
Copyright (C) 1999, 2000, 2001, 2002 Mission Critical Linux, Inc.
This program is free software, covered by the GNU General Public License,
and you are welcome to change it and/or distribute copies of it under
certain conditions. Enter "help copying" to see the conditions.
This program has absolutely no warranty. Enter "help warranty" for
details.
GNU gdb (GDB) 7.6
Copyright (C) 2013 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law. Type "show copying"
and "show warranty" for details.
This GDB was configured as "x86_64-unknown-linux-gnu"...
please wait... (gathering module symbol data)
WARNING: cannot access vmalloc'd module memory
KERNEL: vmlinux-2.6.9-42.0.2.ELsmp.gz
DUMPFILE: vmcore
CPUS: 8
DATE: Tue Nov 21 19:14:17 2006
UPTIME: 6 days, 01:23:25
LOAD AVERAGE: 24.34, 7.89, 4.46
TASKS: 865
NODENAME: lonrs00268
RELEASE: 2.6.9-42.0.2.ELsmp
VERSION: #1 SMP Thu Aug 17 17:57:31 EDT 2006
MACHINE: x86_64 (2199 Mhz)
MEMORY: 16 GB
PANIC: "Kernel BUG at panic:75"
PID: 20046
COMMAND: "oracle"
TASK: 101c6b047f0 [THREAD_INFO: 101a428a000]
CPU: 7
STATE: TASK_RUNNING (NMI)
crash>
If I run the session with "crash -d4 vmlinux-2.6.9-42.0.2.ELsmp.gz vmcore",
you can see that it it reads a "pud page", but then fails:
...
please wait... (gathering module symbol data)module: ffffffffa0634180
<readmem: ffffffffa0634180, KVADDR, "module struct", 1408, (ROE|Q),
f73780>
<readmem: 4f8000, PHYSADDR, "pud page", 4096, (FOE), 2080b40>
<read_diskdump: addr: 4f8000 paddr: 4f8000 cnt: 4096>
crash: invalid kernel virtual address: ffffffffa0634180 type: "module
struct"
WARNING: cannot access vmalloc'd module memory
...
Without the patch, the module virtual address translation succeeds:
...
please wait... (gathering module symbol data)module: ffffffffa0634180
<readmem: ffffffffa0634180, KVADDR, "module struct", 1408, (ROE|Q),
f705e0>
<readmem: 103000, PHYSADDR, "pgd page", 4096, (FOE), 25d7b50>
<read_diskdump: addr: 103000 paddr: 103000 cnt: 4096>
<readmem: 105000, PHYSADDR, "pmd page", 4096, (FOE), 25d8b60>
<read_diskdump: addr: 105000 paddr: 105000 cnt: 4096>
<readmem: d9bfb0000, PHYSADDR, "page table", 4096, (FOE), 25d9b70>
<read_diskdump: addr: d9bfb0000 paddr: d9bfb0000 cnt: 4096>
<read_diskdump: addr: ffffffffa0634180 paddr: d9bfb3180 cnt: 1408>
...
So it appears to be reading from the wrong starting page table location,
i.e., from "pud page 4f8000" instead of "pgd page 103000".
Also, several Xen kernels failed with segmentation violations during
session initialization. They all fail here in x86_64_xendump_load_page(),
when "*pgd" gets referenced:
static char *
x86_64_xendump_load_page(ulong kvaddr, struct xendump_data *xd)
{
ulong mfn;
ulong *pgd, *pud, *pmd, *ptep;
pgd = ((ulong *)machdep->pgd) + pgd_index(kvaddr);
mfn = ((*pgd) & PHYSICAL_PAGE_MASK) >> PAGESHIFT();
^^^^
Here is the relevant part of the gdb trace of a 2.6.18-based xen
kernel:
Program terminated with signal 11, Segmentation fault.
#0 0x0000000000502748 in x86_64_xendump_load_page
(kvaddr=kvaddr@entry=18446744071568498888, xd=0xf521a0 <xendump_data>,
xd=0xf521a0 <xendump_data>) at x86_64.c:7003
7003 mfn = ((*pgd) & PHYSICAL_PAGE_MASK) >> PAGESHIFT();
Missing separate debuginfos, use: debuginfo-install glibc-2.17-157.el7.x86_64
libgcc-4.8.5-11.el7.x86_64 libstdc++-4.8.5-11.el7.x86_64 lzo-2.06-8.el7.x86_64
ncurses-libs-5.9-13.20130511.el7.x86_64 snappy-1.1.0-3.el7.x86_64
xz-libs-5.2.2-1.el7.x86_64 zlib-1.2.7-17.el7.x86_64
(gdb) bt
#0 0x0000000000502748 in x86_64_xendump_load_page
(kvaddr=kvaddr@entry=18446744071568498888, xd=0xf521a0 <xendump_data>,
xd=0xf521a0 <xendump_data>) at x86_64.c:7003
#1 0x0000000000503191 in x86_64_xendump_p2m_create (xd=0xf521a0 <xendump_data>) at
x86_64.c:6749
#2 0x0000000000565d4e in xc_core_create_pfn_tables () at xendump.c:1258
#3 xc_core_read (addr=<optimized out>, paddr=7080864, cnt=32, bufptr=0xf70f80
<shared_bufs>) at xendump.c:168
#4 read_xendump (fd=<optimized out>, bufptr=0xf70f80 <shared_bufs>, cnt=32,
addr=<optimized out>, paddr=7080864) at xendump.c:836
#5 0x000000000047b038 in readmem (addr=18446744071569148832, memtype=memtype@entry=1,
buffer=buffer@entry=0xf70f80 <shared_bufs>,
size=size@entry=32, type=type@entry=0x94dcc3 "possible",
error_handle=error_handle@entry=2) at memory.c:2233
#6 0x00000000004ea33e in cpu_maps_init () at kernel.c:903
#7 kernel_init () at kernel.c:118
#8 0x0000000000467e5a in main_loop () at main.c:768
#9 0x000000000069dad3 in captured_command_loop (data=data@entry=0x0) at main.c:258
#10 0x000000000069c37a in catch_errors (func=func@entry=0x69dac0
<captured_command_loop>, func_args=func_args@entry=0x0,
errstring=errstring@entry=0x8e713f "", mask=mask@entry=6) at
exceptions.c:557
#11 0x000000000069ea66 in captured_main (data=data@entry=0x7ffd637c92a0) at main.c:1064
#12 0x000000000069c37a in catch_errors (func=func@entry=0x69dda0 <captured_main>,
func_args=func_args@entry=0x7ffd637c92a0,
errstring=errstring@entry=0x8e713f "", mask=mask@entry=6) at
exceptions.c:557
#13 0x000000000069edc7 in gdb_main (args=0x7ffd637c92a0) at main.c:1079
#14 gdb_main_entry (argc=<optimized out>, argv=argv@entry=0x7ffd637c9408) at
main.c:1099
#15 0x00000000004f0604 in gdb_main_loop (argc=<optimized out>, argc@entry=3,
argv=argv@entry=0x7ffd637c9408) at gdb_interface.c:76
#16 0x00000000004662c5 in main (argc=3, argv=0x7ffd637c9408) at main.c:707
(gdb) p pgd
$1 = (ulong *) 0xfffffffc054f4210
(gdb)
I haven't investigated further, but in all of the xen cases, the
value of "pgd" above was a kernel virtual address as shown in the
example above.
However, without the patch, the function looks like this, and with
my debug printf of "pml4", the address is a user-space address as
expected:
static char *
x86_64_xendump_load_page(ulong kvaddr, struct xendump_data *xd)
{
ulong mfn;
ulong *pml4, *pgd, *pmd, *ptep;
pml4 = ((ulong *)machdep->machspec->pml4) + pml4_index(kvaddr);
mfn = ((*pml4) & PHYSICAL_PAGE_MASK) >> PAGESHIFT();
fprintf(fp, "x86_64_xendump_load_page: pml4: %lx\n", pml4);
...
So for example, with the debug statement, I see this:
# crash vmlinux-2.6.18-1.2714.el5xen.gz xguest-crashdump
crash 7.2.1rc26
Copyright (C) 2002-2017 Red Hat, Inc.
Copyright (C) 2004, 2005, 2006, 2010 IBM Corporation
Copyright (C) 1999-2006 Hewlett-Packard Co
Copyright (C) 2005, 2006, 2011, 2012 Fujitsu Limited
Copyright (C) 2006, 2007 VA Linux Systems Japan K.K.
Copyright (C) 2005, 2011 NEC Corporation
Copyright (C) 1999, 2002, 2007 Silicon Graphics, Inc.
Copyright (C) 1999, 2000, 2001, 2002 Mission Critical Linux, Inc.
This program is free software, covered by the GNU General Public License,
and you are welcome to change it and/or distribute copies of it under
certain conditions. Enter "help copying" to see the conditions.
This program has absolutely no warranty. Enter "help warranty" for
details.
GNU gdb (GDB) 7.6
Copyright (C) 2013 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law. Type "show copying"
and "show warranty" for details.
This GDB was configured as "x86_64-unknown-linux-gnu"...
x86_64_xendump_load_page: pml4: 25d6c08
x86_64_xendump_load_page: pml4: 25d6c08
KERNEL: vmlinux-2.6.18-1.2714.el5xen.gz
DUMPFILE: xguest-crashdump
...
In a private email, I will send you a pointer to where I have temporarily
stored the 2 vmlinux/vmcore pairs shown above. I'm thinking that it will
probably be fairly easy for you to figure out what's happening in both cases.
Yes, I saw it! Thanks you very much :-)
Thanks,
dou.
Again, I very much appreciate the work you have undertaken here.
Thanks,
Dave
2441
days inactive
2442
days old
devel@lists.crash-utility.osci.io
8 comments
2 participants
participants (2)
-
Dave Anderson -
Dou Liyang