libdrgn: x86_64: avoid recursive address translation for swapper_pg_dir

Most core dumps contain some virtual address mappings: usually at a
minimum, the kernel's direct map is represented in ELF vmcores via a
segment. So normally, drgn can rely on the vmcore to read the virtual
address of swapper_pg_dir. However, some vmcores only contain physical
address information, so when drgn reads memory at swapper_pg_dir, it
needs to first translate that address, thus causing a recursive
translation error like below:

>>> prog["slab_caches"]
Traceback (most recent call last):
  File "<console>", line 1, in <module>
  File "/home/stepbren/repos/drgn/drgn/cli.py", line 141, in _displayhook
    text = value.format_(columns=shutil.get_terminal_size((0, 0)).columns)
_drgn.FaultError: recursive address translation; page table may be missing from core dump: 0xffffffff9662aff8

Debuggers like crash, as well as libkdumpfile, contain fallback code
which can translate swapper_pg_dir in order to bootstrap this address
translation. In fact, the above error does not occur in drgn when using
libkdumpfile. So, let's add this fallback case to drgn as well. Other
architectures will need to have equivalent support added.

Co-authored-by: Illia Ostapyshyn <[email protected]>
Signed-off-by: Stephen Brennan <[email protected]>

libdrgn/arch_x86_64.c

@@ -18,6 +18,10 @@
 
 #include "arch_x86_64_defs.inc"
 
+// This is __START_KERNEL_map from the Linux kernel. It has never been modified
+// since the x86_64 architecture was introduced.
+#define START_KERNEL_MAP UINT64_C(0xffffffff80000000)
+
 static const struct drgn_cfi_row default_dwarf_cfi_row_x86_64 = DRGN_CFI_ROW(
 	// The System V psABI defines the CFA as the value of rsp in the calling
 	// frame.
@@ -616,7 +620,18 @@ linux_kernel_pgtable_iterator_next_x86_64(struct drgn_program *prog,
 	for (;; level--) {
 		uint64_t table;
 		bool table_physical;
-		if (level == levels) {
+		if (level == levels && prog->vmcoreinfo.phys_base &&
+		    it->it.pgtable == prog->vmcoreinfo.swapper_pg_dir) {
+			// Avoid recursive address translation on swapper_pg_dir by
+			// directly resolving to a physical address. Don't do
+			// this if phys_base is 0, since that likely means it
+			// was not present in the vmcoreinfo. It has been
+			// present since Linux kernel commit 401721ecd1dc
+			// ("kexec: export the value of phys_base instead of
+			// symbol address") (in v4.10).
+			table = it->it.pgtable + prog->vmcoreinfo.phys_base - START_KERNEL_MAP;
+			table_physical = true;
+		} else if (level == levels) {
 			table = it->it.pgtable;
 			table_physical = false;
 		} else {

libdrgn/drgn_program_parse_vmcoreinfo.inc.strswitch

@@ -80,6 +80,14 @@ struct drgn_error *drgn_program_parse_vmcoreinfo(struct drgn_program *prog,
 			prog->vmcoreinfo.pgtable_l5_enabled = tmp;
 			break;
 		}
+		@case "NUMBER(phys_base)"@
+		{
+			err = parse_vmcoreinfo_u64(value, newline, 0,
+						   &prog->vmcoreinfo.phys_base);;
+			if (err)
+				return err;
+			break;
+		}
 		@case "NUMBER(KERNELPACMASK)"@
 			err = parse_vmcoreinfo_u64(value, newline, 16,
 						   &prog->aarch64_insn_pac_mask);

libdrgn/program.h

@@ -167,6 +167,8 @@ struct drgn_program {
 		uint64_t mem_section_length;
 		/** VA_BITS on AArch64. */
 		uint64_t va_bits;
+		/** phys_base on x86_64 */
+		uint64_t phys_base;
 		/** Whether 5-level paging was enabled on x86-64. */
 		bool pgtable_l5_enabled;
 		/** PAGE_SHIFT of the kernel (derived from PAGE_SIZE). */