x86/pvh: Support relocating dom0 kernel
Subsystems (3)
v6 cover x86/pvh: Support relocating dom0 kernel
Xen tries to load a PVH dom0 kernel at the fixed guest physical address from the elf headers. For Linux, this defaults to 0x1000000 (16MB), but it can be configured.
Unfortunately there exist firmwares that have reserved regions at this address, so Xen fails to load the dom0 kernel since it's not RAM.
The other issue is that the Linux PVH entry point is not position-independent. It expects to run at the compiled CONFIG_PHYSICAL_ADDRESS.
This patch set expands the PVH dom0 builder to try to relocate the kernel if needed and possible. XEN_ELFNOTE_PHYS32_RELOC is added for kernels to indicate they are relocatable and their acceptable address range and alignment.
v6:
Choose the alignment from the Note if specified, or the Maximum PHDR p_align value if greater than PAGE_SIZE. Otherwise, it falls back to the default 2MB.
Patches from v5 commited:853c71dfbf xen/elfnote: Specify ELF Notes are x86-specific 7d8c9b4e8d libelf: Expand ELF note printing 8802230bfa Revert "xen/x86: bzImage parse kernel_alignment"
The first and second patches move MB/GB() to common-macros.h.
The third patch stores the maximum p_align value from the ELF PHDRs.
The fourth patch expands the pvh dom0 kernel placement code.
I'll post an additional patch showing the Linux changes to make PVH relocatable.
Jason Andryuk (4):
tools/init-xenstore-domain: Replace variable MB() usage
tools: Move MB/GB() to common-macros.h
libelf: Store maximum PHDR p_align
x86/PVH: Support relocatable dom0 kernels
tools/firmware/hvmloader/util.h | 3 - tools/helpers/init-xenstore-domain.c | 11 ++- tools/include/xen-tools/common-macros.h | 4 + tools/libs/light/libxl_internal.h | 4 - xen/arch/x86/hvm/dom0_build.c | 108 ++++++++++++++++++++++++ xen/common/libelf/libelf-dominfo.c | 35 ++++++++ xen/common/libelf/libelf-loader.c | 15 +++- xen/common/libelf/libelf-private.h | 1 + xen/include/public/elfnote.h | 16 +++- xen/include/xen/libelf.h | 5 ++ 10 files changed, 186 insertions(+), 16 deletions(-)
Signature · 1 line
2.44.0
PATCH v6 1/4 tools/init-xenstore-domain: Replace variable MB() usage
The local MB() & GB() macros will be replaced with a common implementation, but those only work with constant values. Introduce a static inline mb_to_bytes() in place of the MB() macro to convert the variable values.
---
v4:
New
--- tools/helpers/init-xenstore-domain.c | 10 +++++++--- 1 file changed, 7 insertions(+), 3 deletions(-)
tools/helpers/init-xenstore-domain.c +7 -3
@@ -20,7 +20,6 @@#include "init-dom-json.h" #define LAPIC_BASE_ADDRESS 0xfee00000UL -#define MB (x) ((uint64_t)x << 20)#define GB (x) ((uint64_t)x << 30)static uint32_t domid = ~0 ;
@@ -36,6 +35,11 @@ static xc_evtchn_port_or_error_t console_evtchn;static xentoollog_level minmsglevel =XTL_PROGRESS ;static void *logger; +static inline uint64_t mb_to_bytes (int mem) +{ +return (uint64_t )mem <<20 ; +} +static struct option options[] = { {"kernel" ,1 ,NULL ,'k' }, {"memory" ,1 ,NULL ,'m' },
@@ -76,8 +80,8 @@ static int build(xc_interface *xch)int rv, xs_fd;struct xc_dom_image *dom =NULL ;int limit_kb = (maxmem ? : memory) *1024 +X86_HVM_NR_SPECIAL_PAGES *4 ; -uint64_t mem_size =MB (memory); -uint64_t max_size =MB (maxmem ? : memory); +uint64_t mem_size =mb_to_bytes (memory); +uint64_t max_size =mb_to_bytes (maxmem ? : memory);struct e820entry e820[3 ];struct xen_domctl_createdomain config = { .ssidref =SECINITSID_DOMU ,
Signature · 1 line
2.44.0
Re: [PATCH v6 1/4] tools/init-xenstore-domain: Replace variable MB() usage
Quoted 2 lines · Jason Andryuk
The local
MB()&GB()macros will be replaced with a common implementation, but those only work with constant values. Introduce a
By the way, this is not true, the macro introduce in the following patch ("tools: Move MB/GB() to common-macros.h") works (compiler doesn't complain) if you do something like MB(maxmem+0) ;-).
Quoted 28 lines
static inline
mb_to_bytes()in place of theMB()macro to convert the variable values.tools/helpers/init-xenstore-domain.c +5
-1@@ -20,7 +20,6 @@#include "init-dom-json.h" #define LAPIC_BASE_ADDRESS 0xfee00000UL -#define MB (x) ((uint64_t)x << 20)#define GB (x) ((uint64_t)x << 30)static uint32_t domid = ~0 ;@@ -36,6 +35,11 @@ static xc_evtchn_port_or_error_t console_evtchn;static xentoollog_level minmsglevel =XTL_PROGRESS ;static void *logger; +static inline uint64_t mb_to_bytes (int mem) +{ +return (uint64_t )mem <<20 ; +} +static struct option options[] = { {"kernel" ,1 ,NULL ,'k' }, {"memory" ,1 ,NULL ,'m' },
tools/helpers/init-xenstore-domain.c static int build(xc_interface *xch) · line 87
Quoted 10 lines
tools/helpers/init-xenstore-domain.c +2
-2@@ -76,8 +80,8 @@ static int build(xc_interface *xch)int rv, xs_fd;struct xc_dom_image *dom =NULL ;int limit_kb = (maxmem ? : memory) *1024 +X86_HVM_NR_SPECIAL_PAGES *4 ; -uint64_t mem_size =MB (memory); -uint64_t max_size =MB (maxmem ? : memory); +uint64_t mem_size =mb_to_bytes (memory); +uint64_t max_size =mb_to_bytes (maxmem ? : memory);struct e820entry e820[3 ];struct xen_domctl_createdomain config = { .ssidref =SECINITSID_DOMU ,
Sorry, just notice they were more versions of the series, so repeating here:
Looks like this change actually fix a bug. When maxmem is set, we would have "max_size = maxmem", otherwise, it would be "max_size = memory << 20"
The macro should have been written as
#define MB(x) ((uint64_t)(x) << 20)
This patch could be backported to 4.17 and later, with:
Anyway, this patch on it's own looks fine:
Signature · 1 line
Anthony PERARD
Re: [PATCH v6 1/4] tools/init-xenstore-domain: Replace variable MB() usage
Quoted 7 lines · Anthony PERARD
Quoted 2 lines · Jason Andryuk
The local
MB()&GB()macros will be replaced with a common implementation, but those only work with constant values. Introduce aBy the way, this is not true, the macro introduce in the following patch ("tools: Move MB/GB() to
common-macros.h") works (compiler doesn't complain) if you do something like MB(maxmem+0) ;-).
Well, I think I wrote the wrong word for the description. The old Macro might have been okay, but the new one as (_AC(_mb, ULL) << 20) fails for MB(memory):
init-xenstore-domain.c: In function ‘build’:
init-xenstore-domain.c:82:28: error: ‘memoryULL’ undeclared (first use in this function); did you mean ‘memory’?
82 | uint64_t mem_size = MB(memory);
| ^~~~~~
I should have written "only work for numeric values."
Quoted 46 lines
Quoted 28 lines
static inline
mb_to_bytes()in place of theMB()macro to convert the variable values.tools/helpers/init-xenstore-domain.c +5
-1@@ -20,7 +20,6 @@#include "init-dom-json.h" #define LAPIC_BASE_ADDRESS 0xfee00000UL -#define MB (x) ((uint64_t)x << 20)#define GB (x) ((uint64_t)x << 30)static uint32_t domid = ~0 ;@@ -36,6 +35,11 @@ static xc_evtchn_port_or_error_t console_evtchn;static xentoollog_level minmsglevel =XTL_PROGRESS ;static void *logger; +static inline uint64_t mb_to_bytes (int mem) +{ +return (uint64_t )mem <<20 ; +} +static struct option options[] = { {"kernel" ,1 ,NULL ,'k' }, {"memory" ,1 ,NULL ,'m' },tools/helpers/init-xenstore-domain.c static int build(xc_interface *xch) · line 87
Quoted 10 lines
tools/helpers/init-xenstore-domain.c +2
-2@@ -76,8 +80,8 @@ static int build(xc_interface *xch)int rv, xs_fd;struct xc_dom_image *dom =NULL ;int limit_kb = (maxmem ? : memory) *1024 +X86_HVM_NR_SPECIAL_PAGES *4 ; -uint64_t mem_size =MB (memory); -uint64_t max_size =MB (maxmem ? : memory); +uint64_t mem_size =mb_to_bytes (memory); +uint64_t max_size =mb_to_bytes (maxmem ? : memory);struct e820entry e820[3 ];struct xen_domctl_createdomain config = { .ssidref =SECINITSID_DOMU ,Sorry, just notice they were more versions of the series, so repeating here:
Looks like this change actually fix a bug. When
maxmemis set, we would have "max_size= maxmem", otherwise, it would be "max_size= memory << 20"
Quoted 5 lines
The macro should have been written as
#define MB(x) ((uint64_t)(x) << 20)This patch could be backported to 4.17 and later, with:
Fixes 134d53f57707 ("tools/init-xenstore-domain: fix memory map for PVH stubdom")
Nice catch!
Quoted 2 lines
Anyway, this patch on it's own looks fine:
Reviewed-by Anthony PERARD <[email protected]>
Thank you.
Regards,
Jason
Re: [PATCH v6 1/4] tools/init-xenstore-domain: Replace variable MB() usage
In reply to Jason Andryuktools/helpers/init-xenstore-domain.c static xc_evtchn_port_or_error_t console_evtchn; · line 41
Quoted 8 lines · Jason Andryuk
tools/helpers/init-xenstore-domain.c +4
-0@@ -36,6 +35,11 @@ static xc_evtchn_port_or_error_t console_evtchn;static xentoollog_level minmsglevel =XTL_PROGRESS ;static void *logger; +static inline uint64_t mb_to_bytes (int mem) +{ +return (uint64_t )mem <<20 ; +}
While committing I noticed tab indentation here, which looked to be in conflict with ...
Quoted 4 lines
tools/helpers/init-xenstore-domain.c +0
-0static struct option options[] = { {"kernel" ,1 ,NULL ,'k' }, {"memory" ,1 ,NULL ,'m' },
tools/helpers/init-xenstore-domain.c static int build(xc_interface *xch) · line 87
Quoted 10 lines
tools/helpers/init-xenstore-domain.c +2
-2@@ -76,8 +80,8 @@ static int build(xc_interface *xch)int rv, xs_fd;struct xc_dom_image *dom =NULL ;int limit_kb = (maxmem ? : memory) *1024 +X86_HVM_NR_SPECIAL_PAGES *4 ; -uint64_t mem_size =MB (memory); -uint64_t max_size =MB (maxmem ? : memory); +uint64_t mem_size =mb_to_bytes (memory); +uint64_t max_size =mb_to_bytes (maxmem ? : memory);struct e820entry e820[3 ];struct xen_domctl_createdomain config = { .ssidref =SECINITSID_DOMU ,
... everything else in the file. Hence I took the liberty to adjust this.
Jan
Re: [PATCH v6 1/4] tools/init-xenstore-domain: Replace variable MB() usage
Quoted 13 lines · Jan Beulich
tools/helpers/init-xenstore-domain.c static xc_evtchn_port_or_error_t console_evtchn; · line 41
Quoted 8 lines · Jason Andryuk
tools/helpers/init-xenstore-domain.c +4
-0@@ -36,6 +35,11 @@ static xc_evtchn_port_or_error_t console_evtchn;static xentoollog_level minmsglevel =XTL_PROGRESS ;static void *logger; +static inline uint64_t mb_to_bytes (int mem) +{ +return (uint64_t )mem <<20 ; +}While committing I noticed tab indentation here, which looked to be in conflict with ...
Quoted 16 lines
Quoted 4 lines
tools/helpers/init-xenstore-domain.c +0
-0static struct option options[] = { {"kernel" ,1 ,NULL ,'k' }, {"memory" ,1 ,NULL ,'m' },tools/helpers/init-xenstore-domain.c static int build(xc_interface *xch) · line 87
Quoted 10 lines
tools/helpers/init-xenstore-domain.c +2
-2@@ -76,8 +80,8 @@ static int build(xc_interface *xch)int rv, xs_fd;struct xc_dom_image *dom =NULL ;int limit_kb = (maxmem ? : memory) *1024 +X86_HVM_NR_SPECIAL_PAGES *4 ; -uint64_t mem_size =MB (memory); -uint64_t max_size =MB (maxmem ? : memory); +uint64_t mem_size =mb_to_bytes (memory); +uint64_t max_size =mb_to_bytes (maxmem ? : memory);struct e820entry e820[3 ];struct xen_domctl_createdomain config = { .ssidref =SECINITSID_DOMU ,... everything else in the file. Hence I took the liberty to adjust this.
Thank you for catching that and fixing it up.
-Jason
PATCH v6 2/4 tools: Move MB/GB() to common-macros.h
In reply to Jason AndryukConsolidate to a single set of common macros for tools.
MB() will gain another use in libelf, so this movement makes it available.
---
v4:
New
v5:
Add Jan's R-b & Req-b
--- tools/firmware/hvmloader/util.h | 3 --- tools/helpers/init-xenstore-domain.c | 1 - tools/include/xen-tools/common-macros.h | 4 ++++ tools/libs/light/libxl_internal.h | 4 ---- 4 files changed, 4 insertions(+), 8 deletions(-)
tools/firmware/hvmloader/util.h +0 -3
@@ -38,9 +38,6 @@ void __bug(const char *file, int line) __attribute__((noreturn));#define BUG () __bug(__FILE__, __LINE__)#define BUG_ON (p) do { if (p) BUG(); } while (0) -#define MB (mb) (mb##ULL << 20) -#define GB (gb) (gb##ULL << 30) -static inline int test_bit (unsigned int b,const void *p) {return !!(((const uint8_t *)p)[b>>3 ] & (1u <<(b&7 )));
tools/helpers/init-xenstore-domain.c +0 -1
@@ -20,7 +20,6 @@#include "init-dom-json.h" #define LAPIC_BASE_ADDRESS 0xfee00000UL -#define GB (x) ((uint64_t)x << 30)static uint32_t domid = ~0 ;static char *kernel;
tools/include/xen-tools/common-macros.h +4 -0
@@ -91,6 +91,10 @@#define __AC (X ,Y ) (X ## Y)#define _AC (X ,Y ) __AC(X, Y) +/* Size macros. */ +#define MB (_mb) (_AC(_mb, ULL) << 20) +#define GB (_gb) (_AC(_gb, ULL) << 30) +#define get_unaligned_t (type, ptr) ({ \ const struct { type x; } __packed *ptr_ = (typeof(ptr_))(ptr); \ ptr_->x; \
tools/libs/light/libxl_internal.h +0 -4
@@ -126,10 +126,6 @@#define PVSHIM_BASENAME "xen-shim"#define PVSHIM_CMDLINE "pv-shim console=xen,pv" -/* Size macros. */ -#define MB (_mb) (_AC(_mb, ULL) << 20) -#define GB (_gb) (_AC(_gb, ULL) << 30) -#define DIV_ROUNDUP (n, d) (((n) + (d) - 1) / (d))#define LIBXL__LOGGING_ENABLED
Signature · 1 line
2.44.0
Re: [PATCH v6 2/4] tools: Move MB/GB() to common-macros.h
Quoted 9 lines · Jason Andryuk
Consolidate to a single set of common macros for tools.
MB()will gain another use in libelf, so this movement makes it available.Requested-by Jan Beulich <[email protected]>Signed-off-by Jason Andryuk <[email protected]>Reviewed-by Jan Beulich <[email protected]>---
So, this patch fixes potential use issues with the macros in hvmloader and init-xenstore-domain. While it's not perfect, it still better.
I'll try "MB(memory + 0)" with the different macros:
tools/firmware/hvmloader/util.h void __bug(const char *file, int line) __attribute__((noreturn)); · line 42
Quoted 10 lines
tools/firmware/hvmloader/util.h +0
-2@@ -38,9 +38,6 @@ void __bug(const char *file, int line) __attribute__((noreturn));#define BUG () __bug(__FILE__, __LINE__)#define BUG_ON (p) do { if (p) BUG(); } while (0) -#define MB (mb) (mb##ULL << 20) -#define GB (gb) (gb##ULL << 30)
With this change we have this change for MB(memory + 0) when applied:
- (memory + 0ULL << 20)
+ ((memory + 0ULL) << 20)
tools/helpers/init-xenstore-domain.c line 23
Quoted 12 lines
tools/firmware/hvmloader/util.h +0
-0static inline int test_bit (unsigned int b,const void *p) {return !!(((const uint8_t *)p)[b>>3 ] & (1u <<(b&7 )));tools/helpers/init-xenstore-domain.c +0
-1@@ -20,7 +20,6 @@#include "init-dom-json.h" #define LAPIC_BASE_ADDRESS 0xfee00000UL -#define GB (x) ((uint64_t)x << 30)
With this change we have this change for GB(memory + 0) when applied:
- ((uint64_t)memory + 0 << 30)
+ ((memory + 0ULL) << 30)
So overall, patch makes things better and less duplication:
Thanks,
Signature · 1 line
Anthony PERARD
PATCH v6 3/4 libelf: Store maximum PHDR p_align
In reply to Jason AndryukWhile parsing the PHDRs, store the maximum p_align value. This may be consulted for moving a PVH image's load address.
---
v6:
New
--- xen/common/libelf/libelf-loader.c | 15 +++++++++++---- xen/include/xen/libelf.h | 1 + 2 files changed, 12 insertions(+), 4 deletions(-)
xen/common/libelf/libelf-loader.c +11 -4
@@ -468,6 +468,7 @@ void elf_parse_binary(struct elf_binary *elf) {ELF_HANDLE_DECL (elf_phdr) phdr;uint64_t low =-1 , high =0 , paddr, memsz; +uint64_t max_align =0 , palign;unsigned i, count; count =elf_phdr_count (elf);
@@ -481,17 +482,23 @@ void elf_parse_binary(struct elf_binary *elf)continue ; paddr =elf_uval (elf, phdr, p_paddr); memsz =elf_uval (elf, phdr, p_memsz); -elf_msg (elf,"ELF: phdr: paddr=%#" PRIx64" memsz=%#" PRIx64"\n" , - paddr, memsz); + palign =elf_uval (elf, phdr, p_align); +elf_msg (elf, +"ELF: phdr: paddr=%#" PRIx64" memsz=%#" PRIx64" palign=%#" PRIx64"\n" , + paddr, memsz, palign);if ( low > paddr ) low = paddr;if ( high < paddr + memsz ) high = paddr + memsz; +if ( max_align < palign ) + max_align = palign; } elf->pstart = low; elf->pend = high; -elf_msg (elf,"ELF: memory: %#" PRIx64" -> %#" PRIx64"\n" , - elf->pstart, elf->pend); + elf->palign = max_align; +elf_msg (elf, +"ELF: memory: %#" PRIx64" -> %#" PRIx64" align:%#" PRIx64"\n" , + elf->pstart, elf->pend, elf->palign); }elf_errorstatus elf_load_binary (struct elf_binary *elf)
xen/include/xen/libelf.h +1 -0
@@ -196,6 +196,7 @@ struct elf_binary {size_t dest_size;uint64_t pstart;uint64_t pend; +uint64_t palign;uint64_t reloc_offset;uint64_t bsd_symtab_pstart;
Signature · 1 line
2.44.0
Re: [PATCH v6 3/4] libelf: Store maximum PHDR p_align
Quoted 26 lines · Jason Andryuk
xen/common/libelf/libelf-loader.c +7
-2@@ -468,6 +468,7 @@ void elf_parse_binary(struct elf_binary *elf) {ELF_HANDLE_DECL (elf_phdr) phdr;uint64_t low =-1 , high =0 , paddr, memsz; +uint64_t max_align =0 , palign;unsigned i, count; count =elf_phdr_count (elf);@@ -481,17 +482,23 @@ void elf_parse_binary(struct elf_binary *elf)continue ; paddr =elf_uval (elf, phdr, p_paddr); memsz =elf_uval (elf, phdr, p_memsz); -elf_msg (elf,"ELF: phdr: paddr=%#" PRIx64" memsz=%#" PRIx64"\n" , - paddr, memsz); + palign =elf_uval (elf, phdr, p_align); +elf_msg (elf, +"ELF: phdr: paddr=%#" PRIx64" memsz=%#" PRIx64" palign=%#" PRIx64"\n" , + paddr, memsz, palign);if ( low > paddr ) low = paddr;if ( high < paddr + memsz ) high = paddr + memsz; +if ( max_align < palign ) + max_align = palign;
xen/common/libelf/libelf-loader.c void elf_parse_binary(struct elf_binary *elf) · line 502
Quoted 10 lines
xen/common/libelf/libelf-loader.c +4
-2} elf->pstart = low; elf->pend = high; -elf_msg (elf,"ELF: memory: %#" PRIx64" -> %#" PRIx64"\n" , - elf->pstart, elf->pend); + elf->palign = max_align; +elf_msg (elf, +"ELF: memory: %#" PRIx64" -> %#" PRIx64" align:%#" PRIx64"\n" , + elf->pstart, elf->pend, elf->palign); }
Hmm, it's just this final logging change which I'm a little concerned by:
Having looked at Linux'es phdr, I noticed that the addresses there aren't necessarily matching the corresponding alignment. Therefore I'm a little concerned that the output here might raise questions when people see seemingly inconsistent values in the log. Could you/we at least make it read like e.g. "align (max): ..."?
Jan
Re: [PATCH v6 3/4] libelf: Store maximum PHDR p_align
Quoted 36 lines · Jan Beulich
Quoted 35 lines · Jason Andryuk
xen/common/libelf/libelf-loader.c +11
-4@@ -468,6 +468,7 @@ void elf_parse_binary(struct elf_binary *elf) {ELF_HANDLE_DECL (elf_phdr) phdr;uint64_t low =-1 , high =0 , paddr, memsz; +uint64_t max_align =0 , palign;unsigned i, count; count =elf_phdr_count (elf);@@ -481,17 +482,23 @@ void elf_parse_binary(struct elf_binary *elf)continue ; paddr =elf_uval (elf, phdr, p_paddr); memsz =elf_uval (elf, phdr, p_memsz); -elf_msg (elf,"ELF: phdr: paddr=%#" PRIx64" memsz=%#" PRIx64"\n" , - paddr, memsz); + palign =elf_uval (elf, phdr, p_align); +elf_msg (elf, +"ELF: phdr: paddr=%#" PRIx64" memsz=%#" PRIx64" palign=%#" PRIx64"\n" , + paddr, memsz, palign);if ( low > paddr ) low = paddr;if ( high < paddr + memsz ) high = paddr + memsz; +if ( max_align < palign ) + max_align = palign; } elf->pstart = low; elf->pend = high; -elf_msg (elf,"ELF: memory: %#" PRIx64" -> %#" PRIx64"\n" , - elf->pstart, elf->pend); + elf->palign = max_align; +elf_msg (elf, +"ELF: memory: %#" PRIx64" -> %#" PRIx64" align:%#" PRIx64"\n" , + elf->pstart, elf->pend, elf->palign);
Quoted 8 lines
xen/common/libelf/libelf-loader.c void elf_parse_binary(struct elf_binary *elf) · line 502
Quoted 1 line
xen/common/libelf/libelf-loader.c +0
-0}Hmm, it's just this final logging change which I'm a little concerned by:
Having looked at Linux'es phdr, I noticed that the addresses there aren't necessarily matching the corresponding alignment. Therefore I'm a little concerned that the output here might raise questions when people see seemingly inconsistent values in the log. Could you/we at least make it read like e.g. "align (max): ..."?
max_align?
Looking at my test vmlinux, it looks like PHDR 0 (.text) and PHDR 1 (.data) are aligned. It's the PHDR 2 (.init) that isn't aligned. Is that what you see?
This line is already printing the min and max across all the PHDRs, so it would only look confusing if the start didn't match the align.
I'm not sure how useful it is to print the alignment, and I considered not printing it. It's only used with PVH Dom0 right now, so it's not relevant in most cases.
Regards,
Jason
Re: [PATCH v6 3/4] libelf: Store maximum PHDR p_align
Quoted 52 lines · Jason Andryuk
Quoted 36 lines · Jan Beulich
Quoted 35 lines · Jason Andryuk
xen/common/libelf/libelf-loader.c +11
-4@@ -468,6 +468,7 @@ void elf_parse_binary(struct elf_binary *elf) {ELF_HANDLE_DECL (elf_phdr) phdr;uint64_t low =-1 , high =0 , paddr, memsz; +uint64_t max_align =0 , palign;unsigned i, count; count =elf_phdr_count (elf);@@ -481,17 +482,23 @@ void elf_parse_binary(struct elf_binary *elf)continue ; paddr =elf_uval (elf, phdr, p_paddr); memsz =elf_uval (elf, phdr, p_memsz); -elf_msg (elf,"ELF: phdr: paddr=%#" PRIx64" memsz=%#" PRIx64"\n" , - paddr, memsz); + palign =elf_uval (elf, phdr, p_align); +elf_msg (elf, +"ELF: phdr: paddr=%#" PRIx64" memsz=%#" PRIx64" palign=%#" PRIx64"\n" , + paddr, memsz, palign);if ( low > paddr ) low = paddr;if ( high < paddr + memsz ) high = paddr + memsz; +if ( max_align < palign ) + max_align = palign; } elf->pstart = low; elf->pend = high; -elf_msg (elf,"ELF: memory: %#" PRIx64" -> %#" PRIx64"\n" , - elf->pstart, elf->pend); + elf->palign = max_align; +elf_msg (elf, +"ELF: memory: %#" PRIx64" -> %#" PRIx64" align:%#" PRIx64"\n" , + elf->pstart, elf->pend, elf->palign);Quoted 8 lines
xen/common/libelf/libelf-loader.c void elf_parse_binary(struct elf_binary *elf) · line 502
Quoted 1 line
xen/common/libelf/libelf-loader.c +0
-0}Hmm, it's just this final logging change which I'm a little concerned by:
Having looked at Linux'es phdr, I noticed that the addresses there aren't necessarily matching the corresponding alignment. Therefore I'm a little concerned that the output here might raise questions when people see seemingly inconsistent values in the log. Could you/we at least make it read like e.g. "align (max): ..."?
max_align?Looking at my test vmlinux, it looks like PHDR 0 (.text) and PHDR 1 (.data) are aligned. It's the PHDR 2 (.init) that isn't aligned. Is that what you see?
Quoted 6 lines
This line is already printing the min and max across all the PHDRs, so it would only look confusing if the start didn't match the align.
I'm not sure how useful it is to print the alignment, and I considered not printing it. It's only used with PVH Dom0 right now, so it's not relevant in most cases.
Well, yes, not printing the alignment would be fine with me. I just didn't suggest that as an alternative since I was assuming you put its printing there for a reason.
Jan
Re: [PATCH v6 3/4] libelf: Store maximum PHDR p_align
Quoted 56 lines · Jan Beulich
Quoted 55 lines · Jason Andryuk
Quoted 36 lines · Jan Beulich
Quoted 35 lines · Jason Andryuk
xen/common/libelf/libelf-loader.c +11
-4@@ -468,6 +468,7 @@ void elf_parse_binary(struct elf_binary *elf) {ELF_HANDLE_DECL (elf_phdr) phdr;uint64_t low =-1 , high =0 , paddr, memsz; +uint64_t max_align =0 , palign;unsigned i, count; count =elf_phdr_count (elf);@@ -481,17 +482,23 @@ void elf_parse_binary(struct elf_binary *elf)continue ; paddr =elf_uval (elf, phdr, p_paddr); memsz =elf_uval (elf, phdr, p_memsz); -elf_msg (elf,"ELF: phdr: paddr=%#" PRIx64" memsz=%#" PRIx64"\n" , - paddr, memsz); + palign =elf_uval (elf, phdr, p_align); +elf_msg (elf, +"ELF: phdr: paddr=%#" PRIx64" memsz=%#" PRIx64" palign=%#" PRIx64"\n" , + paddr, memsz, palign);if ( low > paddr ) low = paddr;if ( high < paddr + memsz ) high = paddr + memsz; +if ( max_align < palign ) + max_align = palign; } elf->pstart = low; elf->pend = high; -elf_msg (elf,"ELF: memory: %#" PRIx64" -> %#" PRIx64"\n" , - elf->pstart, elf->pend); + elf->palign = max_align; +elf_msg (elf, +"ELF: memory: %#" PRIx64" -> %#" PRIx64" align:%#" PRIx64"\n" , + elf->pstart, elf->pend, elf->palign);Quoted 8 lines
xen/common/libelf/libelf-loader.c void elf_parse_binary(struct elf_binary *elf) · line 502
Quoted 1 line
xen/common/libelf/libelf-loader.c +0
-0}Hmm, it's just this final logging change which I'm a little concerned by:
Having looked at Linux'es phdr, I noticed that the addresses there aren't necessarily matching the corresponding alignment. Therefore I'm a little concerned that the output here might raise questions when people see seemingly inconsistent values in the log. Could you/we at least make it read like e.g. "align (max): ..."?
max_align?Looking at my test vmlinux, it looks like PHDR 0 (.text) and PHDR 1 (.data) are aligned. It's the PHDR 2 (.init) that isn't aligned. Is that what you see?
This line is already printing the min and max across all the PHDRs, so it would only look confusing if the start didn't match the align.
Quoted 7 lines
Quoted 3 lines
I'm not sure how useful it is to print the alignment, and I considered not printing it. It's only used with PVH Dom0 right now, so it's not relevant in most cases.
Well, yes, not printing the alignment would be fine with me. I just didn't suggest that as an alternative since I was assuming you put its printing there for a reason.
I added it to make the information available, which might be useful if there was a problem finding a load location. Since it's usually not useful information, I'll just drop it.
Thanks,
Jason
PATCH v6 4/4 x86/PVH: Support relocatable dom0 kernels
In reply to Jason AndryukXen tries to load a PVH dom0 kernel at the fixed guest physical address from the elf headers. For Linux, this defaults to 0x1000000 (16MB), but it can be configured.
Unfortunately there exist firmwares that have reserved regions at this address, so Xen fails to load the dom0 kernel since it's not RAM.
The PVH entry code is not relocatable - it loads from absolute addresses, which fail when the kernel is loaded at a different address. With a suitably modified kernel, a reloctable entry point is possible.
Add XEN_ELFNOTE_PHYS32_RELOC which specifies optional alignment, minimum, and maximum addresses needed for the kernel. The presence of the NOTE indicates the kernel supports a relocatable entry path.
Change the loading to check for an acceptable load address. If the kernel is relocatable, support finding an alternate load address.
The primary motivation for an explicit align field is that Linux has a configurable CONFIG_PHYSICAL_ALIGN field. This value is present in the bzImage setup header, but not the ELF program headers p_align, which report 2MB even when CONFIG_PHYSICAL_ALIGN is greater. Since a kernel is only considered relocatable if the PHYS32_RELOC elf note is present, the alignment contraints can just be specified within the note instead of searching for an alignment value via a heuristic.
Load alignment uses the PHYS32_RELOC note value if specified. Otherwise, the maxmum ELF PHDR p_align value is selected if greater than or equal to PAGE_SIZE. Finally, the fallback default is 2MB.
libelf-private.h includes common-macros.h to satisfy the fuzzer build.
---
ELF Note printing looks like:
(XEN) ELF: note: PHYS32_RELOC align: 0x200000 min: 0x1000000 max: 0x3fffffff
v2:
Use elfnote for min, max & align - use 64bit values. Print original and relocated memory addresses
Use check_and_adjust_load_address() name
Return relocated base instead of offset
Use PAGE_ALIGN
Don't load above max_phys (expected to be 4GB in kernel elf note) Use single line comments
Exit check_load_address loop earlier
Add __init to find_kernel_memory()
v3:
Remove kernel_start/end page rounding Change loop comment to rely on a sorted memory map. Reorder E820_RAM check first
Use %p for dest_base
Use PRIpaddr
Use 32bit phys_min/max/align
Consolidate to if ( x || y ) continue
Use max_t
Add parms->phys_reloc
Use common "%pd kernel: " prefix for messages Re-order phys_entry assignment
Print range ends inclusively
Remove extra "Unable to load kernel" message s/PVH_RELOCATION/PHYS32_RELOC/
Make PHYS32_RELOC contents optional
Use 2MB default alignment
Update ELF Note comment
Update XEN_ELFNOTE_MAX
v4:
Cast dest_base to uintptr_t
Use local start variable
Mention e820 requiring adjacent entries merged
Remove extra whitespace
Re-word elfnote comment & mention x86
Use ELFNOTE_NAME
Return -ENOSPC
Use ! instead of == 0
Check kend - 1 to avoid off by one
libelf: Use MB/GB() to define the size.
Use ARCH_PHYS_* defines
v5:
Place kernel in higher memory addresses
Remove stray semicolonsELFNOTE_NAME comment about newline
Make PHYS32_RELOC element order align, min, max
Re-word PHYS32_RELOC comment
Move phys_align next to other bool variables
v6:
Select alignment from, in order, Note, PHDRs, then default
--- xen/arch/x86/hvm/dom0_build.c | 108 +++++++++++++++++++++++++++++ xen/common/libelf/libelf-dominfo.c | 35 ++++++++++ xen/common/libelf/libelf-private.h | 1 + xen/include/public/elfnote.h | 16 ++++- xen/include/xen/libelf.h | 4 ++ 5 files changed, 163 insertions(+), 1 deletion(-)
xen/arch/x86/hvm/dom0_build.c +108 -0
@@ -537,6 +537,111 @@ static paddr_t __init find_memory(return INVALID_PADDR ; } +static bool __init check_load_address ( +const struct domain *d,const struct elf_binary *elf) +{ +paddr_t kernel_start = (uintptr_t )elf->dest_base; +paddr_t kernel_end = kernel_start + elf->dest_size; +unsigned int i; + +/* Relies on a sorted memory map with adjacent entries merged. */ +for ( i =0 ; i < d->arch.nr_e820; i++ ) + { +paddr_t start = d->arch.e820[i].addr; +paddr_t end = start + d->arch.e820[i].size; + +if ( start >= kernel_end ) +return false; + +if ( d->arch.e820[i].type ==E820_RAM && + start <= kernel_start && + end >= kernel_end ) +return true; + } + +return false; +} + +/* Find an e820 RAM region that fits the kernel at a suitable alignment. */ +static paddr_t __init find_kernel_memory ( +const struct domain *d,struct elf_binary *elf, +const struct elf_dom_parms *parms) +{ +paddr_t kernel_size = elf->dest_size; +unsigned int align; +int i; + +if ( parms->phys_align !=UNSET_ADDR32 ) + align = parms->phys_align; +else if ( elf->palign >PAGE_SIZE ) + align = elf->palign; +else + align =MB (2 ); + +/* Search backwards to find the highest address. */ +for ( i = d->arch.nr_e820 -1 ; i >=0 ; i-- ) + { +paddr_t start = d->arch.e820[i].addr; +paddr_t end = start + d->arch.e820[i].size; +paddr_t kstart, kend; + +if ( d->arch.e820[i].type !=E820_RAM || + d->arch.e820[i].size < kernel_size ) +continue ; + +if ( start > parms->phys_max ) +continue ; + +if ( end -1 > parms->phys_max ) + end = parms->phys_max +1 ; + + kstart = (end - kernel_size) & ~(align -1 ); + kend = kstart + kernel_size; + +if ( kstart < parms->phys_min ) +return 0 ; + +if ( kstart >= start && kend <= end ) +return kstart; + } + +return 0 ; +} + +/* Check the kernel load address, and adjust if necessary and possible. */ +static bool __init check_and_adjust_load_address ( +const struct domain *d,struct elf_binary *elf,struct elf_dom_parms *parms) +{ +paddr_t reloc_base; + +if (check_load_address (d, elf) ) +return true; + +if ( !parms->phys_reloc ) + { +printk ("%pd kernel: Address conflict and not relocatable\n" , d); +return false; + } + + reloc_base =find_kernel_memory (d, elf, parms); +if ( !reloc_base ) + { +printk ("%pd kernel: Failed find a load address\n" , d); +return false; + } + +if ( opt_dom0_verbose ) +printk ("%pd kernel: Moving [%p, %p] -> [%" PRIpaddr", %" PRIpaddr"]\n" , d, + elf->dest_base, elf->dest_base + elf->dest_size -1 , + reloc_base, reloc_base + elf->dest_size -1 ); + + parms->phys_entry = reloc_base + + (parms->phys_entry - (uintptr_t )elf->dest_base); + elf->dest_base = (char *)reloc_base; + +return true; +} + static int __init pvh_load_kernel(struct domain *d, const module_t *image, unsigned long image_headroom, module_t *initrd, void *image_base,
@@ -585,6 +690,9 @@ static int __init pvh_load_kernel(struct domain *d, const module_t *image, elf.dest_base = (void *)(parms.virt_kstart - parms.virt_base); elf.dest_size = parms.virt_kend - parms.virt_kstart; +if ( !check_and_adjust_load_address (d, &elf, &parms) ) +return -ENOSPC ; +elf_set_vcpu (&elf, v); rc =elf_load_binary (&elf);if ( rc <0 )
xen/common/libelf/libelf-dominfo.c +35 -0
@@ -17,6 +17,14 @@#include "libelf-private.h" +#if defined(__i386__) || defined(__x86_64__) +#define ARCH_PHYS_MIN_DEFAULT 0 +#define ARCH_PHYS_MAX_DEFAULT (GB(4) - 1) +#else +#define ARCH_PHYS_MIN_DEFAULT 0 +#define ARCH_PHYS_MAX_DEFAULT 0 +#endif +/* ------------------------------------------------------------------------ */ /* xen features */
@@ -125,6 +133,7 @@ elf_errorstatus elf_xen_parse_note(struct elf_binary *elf, [XEN_ELFNOTE_SUSPEND_CANCEL] = { "SUSPEND_CANCEL", ELFNOTE_INT }, [XEN_ELFNOTE_MOD_START_PFN] = { "MOD_START_PFN", ELFNOTE_INT }, [XEN_ELFNOTE_PHYS32_ENTRY] = { "PHYS32_ENTRY", ELFNOTE_INT }, + [XEN_ELFNOTE_PHYS32_RELOC] = { "PHYS32_RELOC", ELFNOTE_NAME }, };/* *INDENT-ON* */
@@ -132,6 +141,7 @@ elf_errorstatus elf_xen_parse_note(struct elf_binary *elf,uint64_t val =0 ;unsigned int i;unsigned type =elf_uval (elf, note, type); +unsigned descsz =elf_uval (elf, note, descsz);if ( (type >=sizeof (note_desc) /sizeof (note_desc[0 ])) || (note_desc[type].name ==NULL ) )
@@ -228,6 +238,27 @@ elf_errorstatus elf_xen_parse_note(struct elf_binary *elf,case XEN_ELFNOTE_PHYS32_ENTRY : parms->phys_entry = val;break ; + +case XEN_ELFNOTE_PHYS32_RELOC : + parms->phys_reloc = true; + +if ( descsz >=4 ) + { + parms->phys_align =elf_note_numeric_array (elf, note,4 ,0 ); +elf_msg (elf," align: %#" PRIx32, parms->phys_align); + } +if ( descsz >=8 ) + { + parms->phys_min =elf_note_numeric_array (elf, note,4 ,1 ); +elf_msg (elf," min: %#" PRIx32, parms->phys_min); + } +if ( descsz >=12 ) + { + parms->phys_max =elf_note_numeric_array (elf, note,4 ,2 ); +elf_msg (elf," max: %#" PRIx32, parms->phys_max); + } + +break ; }if ( note_desc[type].type ==ELFNOTE_NAME )
@@ -543,6 +574,10 @@ elf_errorstatus elf_xen_parse(struct elf_binary *elf, parms->p2m_base =UNSET_ADDR ; parms->elf_paddr_offset =UNSET_ADDR ; parms->phys_entry =UNSET_ADDR32 ; + parms->phys_align =UNSET_ADDR32 ; + parms->phys_min =ARCH_PHYS_MIN_DEFAULT ; + parms->phys_max =ARCH_PHYS_MAX_DEFAULT ; + parms->phys_reloc = false;/* Find and parse elf notes. */ count =elf_phdr_count (elf);
xen/common/libelf/libelf-private.h +1 -0
@@ -71,6 +71,7 @@#endif #include <xen/elfnote.h> #include <xen/libelf/libelf.h> +#include <xen-tools/common-macros.h> #ifndef FUZZ_NO_LIBXC #include "xenctrl.h"
xen/include/public/elfnote.h +15 -1
@@ -196,10 +196,24 @@*/ #define XEN_ELFNOTE_PHYS32_ENTRY 18 +/* +* Physical loading constraints for PVH kernels +* +* The presence of this note indicates the kernel supports relocating itself. +* +* The note may include up to three 32bit values to place constraints on the +* guest physical loading addresses and alignment for a PVH kernel. Values +* are read in the following order: +* - a required start alignment (default 0x200000) +* - a minimum address for the start of the image (default 0) +* - a maximum address for the last byte of the image (default 0xffffffff) +*/ +#define XEN_ELFNOTE_PHYS32_RELOC 19 +/* * The number of the highest elfnote defined. */ -#define XEN_ELFNOTE_MAX XEN_ELFNOTE_PHYS32_ENTRY +#define XEN_ELFNOTE_MAX XEN_ELFNOTE_PHYS32_RELOC/* * System information exported through crash notes.
xen/include/xen/libelf.h +4 -0
@@ -426,6 +426,7 @@ struct elf_dom_parms {enum xen_pae_type pae;bool bsd_symtab;bool unmapped_initrd; +bool phys_reloc;uint64_t virt_base;uint64_t virt_entry;uint64_t virt_hypercall;
@@ -435,6 +436,9 @@ struct elf_dom_parms {uint32_t f_supported[XENFEAT_NR_SUBMAPS ];uint32_t f_required[XENFEAT_NR_SUBMAPS ];uint32_t phys_entry; +uint32_t phys_align; +uint32_t phys_min; +uint32_t phys_max;/* calculated */ uint64_t virt_kstart;
Signature · 1 line
2.44.0
Re: [PATCH v6 4/4] x86/PVH: Support relocatable dom0 kernels
Quoted 2 lines · Jason Andryuk
v6:
Select alignment from, in order, Note, PHDRs, then default
The comment in the public header also needs to reflect this change.
Quoted 18 lines
xen/arch/x86/hvm/dom0_build.c +18
-0+static bool __init check_and_adjust_load_address ( +const struct domain *d,struct elf_binary *elf,struct elf_dom_parms *parms) +{ +paddr_t reloc_base; + +if (check_load_address (d, elf) ) +return true; + +if ( !parms->phys_reloc ) + { +printk ("%pd kernel: Address conflict and not relocatable\n" , d); +return false; + } + + reloc_base =find_kernel_memory (d, elf, parms); +if ( !reloc_base ) + { +printk ("%pd kernel: Failed find a load address\n" , d);
xen/arch/x86/hvm/dom0_build.c static paddr_t __init find_memory( · line 639
Quoted 10 lines
xen/arch/x86/hvm/dom0_build.c +10
-0+return false; + } + +if ( opt_dom0_verbose ) +printk ("%pd kernel: Moving [%p, %p] -> [%" PRIpaddr", %" PRIpaddr"]\n" , d, + elf->dest_base, elf->dest_base + elf->dest_size -1 , + reloc_base, reloc_base + elf->dest_size -1 ); + + parms->phys_entry = reloc_base + + (parms->phys_entry - (uintptr_t )elf->dest_base);
I think this would be easier to read as
parms->phys_entry =
reloc_base + (parms->phys_entry - (uintptr_t)elf->dest_base);Everything else looks good to me now.
Jan
Re: [PATCH v6 4/4] x86/PVH: Support relocatable dom0 kernels
Quoted 5 lines · Jan Beulich
Quoted 2 lines · Jason Andryuk
v6:
Select alignment from, in order, Note, PHDRs, then defaultThe comment in the public header also needs to reflect this change.
Yes. I'll update.
I'm going to tweak the code from elf->palign > PAGE_SIZE to >=, since PAGE_SIZE is a reasonable value to use.
Quoted 31 lines
Quoted 18 lines
xen/arch/x86/hvm/dom0_build.c +18
-0+static bool __init check_and_adjust_load_address ( +const struct domain *d,struct elf_binary *elf,struct elf_dom_parms *parms) +{ +paddr_t reloc_base; + +if (check_load_address (d, elf) ) +return true; + +if ( !parms->phys_reloc ) + { +printk ("%pd kernel: Address conflict and not relocatable\n" , d); +return false; + } + + reloc_base =find_kernel_memory (d, elf, parms); +if ( !reloc_base ) + { +printk ("%pd kernel: Failed find a load address\n" , d);xen/arch/x86/hvm/dom0_build.c static paddr_t __init find_memory( · line 639
Quoted 10 lines
xen/arch/x86/hvm/dom0_build.c +10
-0+return false; + } + +if ( opt_dom0_verbose ) +printk ("%pd kernel: Moving [%p, %p] -> [%" PRIpaddr", %" PRIpaddr"]\n" , d, + elf->dest_base, elf->dest_base + elf->dest_size -1 , + reloc_base, reloc_base + elf->dest_size -1 ); + + parms->phys_entry = reloc_base + + (parms->phys_entry - (uintptr_t )elf->dest_base);I think this would be easier to read as
Quoted 4 lines
parms->phys_entry = reloc_base + (parms->phys_entry - (uintptr_t)elf->dest_base);Everything else looks good to me now.
Sounds good to me.
Thanks,
Jason
PATCH v6 1/1 RFC: x86/pvh: Make Xen PVH entrypoint PIC for x86-64
In reply to Jason AndryukThe Xen PVH entrypoint is 32bit non-PIC code running at a default load address of 0x1000000 (16MB) (CONFIG_PHYSICAL_START). Xen loads the kernel at that physical address inside the PVH container.
When running a PVH Dom0, the system reserved addresses are mapped 1-1 into the PVH container. There exist system firmwares (Coreboot/EDK2) with reserved memory at 16MB. This creates a conflict where the PVH kernel cannot be loaded at that address.
Modify the PVH entrypoint to be position-indepedent to allow flexibility in load address. Only the 64bit entry path is converted. A 32bit kernel is not PIC, so calling into other parts of the kernel, like xen_prepare_pvh() and mk_pgtable_32(), don't work properly when relocated.
Initial PVH entry runs at the physical addresses and then transitions to the identity mapped address. While executing xen_prepare_pvh() calls through pv_ops function pointers transition to the high mapped addresses. Additionally, __va() is called on some hvm_start_info physical addresses, we need the directmap address range is used. So we need to run page tables with all of those ranges mapped.
Modifying init_top_pgt tables ran into issue since startup_64/__startup_64() will modify those page tables again. Use a dedicated set of page tables - pvh_init_top_pgt - for the PVH entry to avoid unwanted interactions.
In xen_pvh_init(), __pa() is called to find the physical address of the hypercall page. Set phys_base temporarily before calling into xen_prepare_pvh(), which calls xen_pvh_init(), and clear it afterwards. __startup_64() assumes phys_base is zero and adds load_delta to it. If phys_base is already set, the calculation results in an incorrect cr3.
TODO: Sync elfnote.h from xen.git commit xxxxxxxxxx when it is commited.
---
Put this out as an example for the Xen modifications
Instead of setting and clearing phys_base, add a dedicated variable? Clearing phys_base is a little weird, but it leaves the kernel more consistent when running non-entry code.
Make __startup_64() exit if phys_base is already set to allow calling multiple times, and use that and init_top_pgt instead of adding additional page tables? That won't work. __startup_64 is 64bit code, and pvh needs to create page tables in 32bit code before it can transition to 64bit long mode. Hence it can't be re-used to relocate page tables.
--- arch/x86/platform/pvh/head.S | 182 +++++++++++++++++++++++++++++--- include/xen/interface/elfnote.h | 16 ++- 2 files changed, 185 insertions(+), 13 deletions(-)
arch/x86/platform/pvh/head.S +170 -12
@@ -50,11 +50,32 @@ #define PVH_CS_SEL (PVH_GDT_ENTRY_CS * 8) #define PVH_DS_SEL (PVH_GDT_ENTRY_DS * 8) +#define rva(x) ((x) - pvh_start_xen) + SYM_CODE_START_LOCAL(pvh_start_xen) UNWIND_HINT_END_OF_STACK cld - lgdt (_pa(gdt)) + /* + * See the comment for startup_32 for more details. We need to + * execute a call to get the execution address to be position + * independent, but we don't have a stack. Save and restore the + * magic field of start_info in ebx, and use that as the stack. + */ + mov (%ebx), %eax + leal 4(%ebx), %esp + ANNOTATE_INTRA_FUNCTION_CALL + call 1f +1: popl %ebp + mov %eax, (%ebx) + subl $ rva(1b), %ebp + movl $0, %esp + + leal rva(gdt)(%ebp), %eax + movl %eax, %ecx + leal rva(gdt_start)(%ebp), %ecx + movl %ecx, 2(%eax) + lgdt (%eax) mov $PVH_DS_SEL,%eax mov %eax,%ds
@@ -62,14 +83,14 @@ SYM_CODE_START_LOCAL(pvh_start_xen) mov %eax,%ss /* Stash hvm_start_info. */ - mov $_pa(pvh_start_info), %edi + leal rva(pvh_start_info)(%ebp), %edi mov %ebx, %esi - mov _pa(pvh_start_info_sz), %ecx + movl rva(pvh_start_info_sz)(%ebp), %ecx shr $2,%ecx rep movsl - mov $_pa(early_stack_end), %esp + leal rva(early_stack_end)(%ebp), %esp /* Enable PAE mode. */ mov %cr4, %eax
@@ -83,29 +104,81 @@ SYM_CODE_START_LOCAL(pvh_start_xen) btsl $_EFER_LME, %eax wrmsr + mov %ebp, %ebx + subl $LOAD_PHYSICAL_ADDR, %ebx /* offset */ + jz .Lpagetable_done + + /* Fixup page-tables for relocation. */ + leal rva(pvh_init_top_pgt)(%ebp), %edi + movl $512, %ecx +2: + testl $_PAGE_PRESENT, 0x00(%edi) + jz 1f + addl %ebx, 0x00(%edi) +1: + addl $8, %edi + decl %ecx + jnz 2b + + /* L3 ident has a single entry. */ + leal rva(pvh_level3_ident_pgt)(%ebp), %edi + addl %ebx, 0x00(%edi) + + leal rva(pvh_level3_kernel_pgt)(%ebp), %edi + addl %ebx, (4096 - 16)(%edi) + addl %ebx, (4096 - 8)(%edi) + + /* pvh_level2_ident_pgt is fine - large pages */ + + /* pvh_level2_kernel_pgt needs adjustment - large pages */ + leal rva(pvh_level2_kernel_pgt)(%ebp), %edi + movl $512, %ecx +2: + testl $_PAGE_PRESENT, 0x00(%edi) + jz 1f + addl %ebx, 0x00(%edi) +1: + addl $8, %edi + decl %ecx + jnz 2b + +.Lpagetable_done: /* Enable pre-constructed page tables. */ - mov $_pa(init_top_pgt), %eax + leal rva(pvh_init_top_pgt)(%ebp), %eax mov %eax, %cr3 mov $(X86_CR0_PG | X86_CR0_PE), %eax mov %eax, %cr0 /* Jump to 64-bit mode. */ - ljmp $PVH_CS_SEL, $_pa(1f) + pushl $PVH_CS_SEL + leal rva(1f)(%ebp), %eax + pushl %eax + lretl /* 64-bit entry point. */ .code64 1: /* Set base address in stack canary descriptor. */ mov $MSR_GS_BASE,%ecx - mov $_pa(canary), %eax + leal rva(canary)(%ebp), %eax xor %edx, %edx wrmsr + /* Calculate load offset from LOAD_PHYSICAL_ADDR and store in + * phys_base. __pa() needs phys_base set to calculate the the + * hypercall page in xen_pvh_init(). */ + movq %rbp, %rbx + subq $LOAD_PHYSICAL_ADDR, %rbx + movq %rbx, phys_base(%rip) call xen_prepare_pvh + /* Clear phys_base. startup_64/__startup_64 will *add* to its value, + so start from 0. */ + xor %rbx, %rbx + movq %rbx, phys_base(%rip) /* startup_64 expects boot_params in %rsi. */ - mov $_pa(pvh_bootparams), %rsi - mov $_pa(startup_64), %rax + lea rva(pvh_bootparams)(%ebp), %rsi + lea rva(startup_64)(%ebp), %rax ANNOTATE_RETPOLINE_SAFE jmp *%rax
@@ -137,13 +210,14 @@ SYM_CODE_START_LOCAL(pvh_start_xen) ljmp $PVH_CS_SEL, $_pa(startup_32) #endif + SYM_CODE_END(pvh_start_xen) .section ".init.data","aw" .balign 8 SYM_DATA_START_LOCAL(gdt) - .word gdt_end - gdt_start - .long _pa(gdt_start) + .word gdt_end - gdt_start - 1 + .long _pa(gdt_start) /* x86-64 will overwrite if relocated. */ .word 0 SYM_DATA_END(gdt) SYM_DATA_START_LOCAL(gdt_start)
@@ -163,5 +237,89 @@ SYM_DATA_START_LOCAL(early_stack) .fill BOOT_STACK_SIZE, 1, 0 SYM_DATA_END_LABEL(early_stack, SYM_L_LOCAL, early_stack_end) +#ifdef CONFIG_X86_64 +/* + * We are not able to switch in one step to the final KERNEL ADDRESS SPACE + * because we need identity-mapped pages. + */ +#define l4_index(x) (((x) >> 39) & 511) +#define pud_index(x) (((x) >> PUD_SHIFT) & (PTRS_PER_PUD-1)) + +L4_PAGE_OFFSET = l4_index(__PAGE_OFFSET_BASE_L4) +L4_START_KERNEL = l4_index(__START_KERNEL_map) +L3_START_KERNEL = pud_index(__START_KERNEL_map) + +#define SYM_DATA_START_PAGE_ALIGNED(name) \ + SYM_START(name, SYM_L_GLOBAL, .balign PAGE_SIZE) + +/* Automate the creation of 1 to 1 mapping pmd entries */ +#define PMDS(START, PERM, COUNT) \ + i = 0 ; \ + .rept (COUNT) ; \ + .quad (START) + (i << PMD_SHIFT) + (PERM) ; \ + i = i + 1 ; \ + .endr + +/* + * Xen PVH needs a set of identity mapped and kernel high mapping + * page tables. pvh_start_xen starts running on the identity mapped + * page tables, but xen_prepare_pvh calls into the high mapping. + * These page tables need to be relocatable and are only used until + * startup_64 transitions to init_top_pgt. + */ +SYM_DATA_START_PAGE_ALIGNED(pvh_init_top_pgt) + .quad pvh_level3_ident_pgt - __START_KERNEL_map + _KERNPG_TABLE_NOENC + .org pvh_init_top_pgt + L4_PAGE_OFFSET*8, 0 + .quad pvh_level3_ident_pgt - __START_KERNEL_map + _KERNPG_TABLE_NOENC + .org pvh_init_top_pgt + L4_START_KERNEL*8, 0 + /* (2^48-(2*1024*1024*1024))/(2^39) = 511 */ + .quad pvh_level3_kernel_pgt - __START_KERNEL_map + _PAGE_TABLE_NOENC +SYM_DATA_END(pvh_init_top_pgt) + +SYM_DATA_START_PAGE_ALIGNED(pvh_level3_ident_pgt) + .quad pvh_level2_ident_pgt - __START_KERNEL_map + _KERNPG_TABLE_NOENC + .fill 511, 8, 0 +SYM_DATA_END(pvh_level3_ident_pgt) +SYM_DATA_START_PAGE_ALIGNED(pvh_level2_ident_pgt) + /* + * Since I easily can, map the first 1G. + * Don't set NX because code runs from these pages. + * + * Note: This sets _PAGE_GLOBAL despite whether + * the CPU supports it or it is enabled. But, + * the CPU should ignore the bit. + */ + PMDS(0, __PAGE_KERNEL_IDENT_LARGE_EXEC, PTRS_PER_PMD) +SYM_DATA_END(pvh_level2_ident_pgt) +SYM_DATA_START_PAGE_ALIGNED(pvh_level3_kernel_pgt) + .fill L3_START_KERNEL,8,0 + /* (2^48-(2*1024*1024*1024)-((2^39)*511))/(2^30) = 510 */ + .quad pvh_level2_kernel_pgt - __START_KERNEL_map + _KERNPG_TABLE_NOENC + .quad 0 /* no fixmap */ +SYM_DATA_END(pvh_level3_kernel_pgt) + +SYM_DATA_START_PAGE_ALIGNED(pvh_level2_kernel_pgt) + /* + * Kernel high mapping. + * + * The kernel code+data+bss must be located below KERNEL_IMAGE_SIZE in + * virtual address space, which is 1 GiB if RANDOMIZE_BASE is enabled, + * 512 MiB otherwise. + * + * (NOTE: after that starts the module area, see MODULES_VADDR.) + * + * This table is eventually used by the kernel during normal runtime. + * Care must be taken to clear out undesired bits later, like _PAGE_RW + * or _PAGE_GLOBAL in some cases. + */ + PMDS(0, __PAGE_KERNEL_LARGE_EXEC, KERNEL_IMAGE_SIZE/PMD_SIZE) +SYM_DATA_END(pvh_level2_kernel_pgt) + + ELFNOTE(Xen, XEN_ELFNOTE_PHYS32_RELOC, + .long CONFIG_PHYSICAL_ALIGN; + .long LOAD_PHYSICAL_ADDR; + .long KERNEL_IMAGE_SIZE - 1) +#endif + ELFNOTE(Xen, XEN_ELFNOTE_PHYS32_ENTRY, - _ASM_PTR (pvh_start_xen - __START_KERNEL_map)) + .long (pvh_start_xen - __START_KERNEL_map))
include/xen/interface/elfnote.h +15 -1
@@ -185,9 +185,23 @@*/ #define XEN_ELFNOTE_PHYS32_ENTRY 18 +/* +* Physical loading constraints for PVH kernels +* +* The presence of this note indicates the kernel supports relocating itself. +* +* The note may include up to three 32bit values to place constraints on the +* guest physical loading addresses and alignment for a PVH kernel. Values +* are read in the following order: +* - a required start alignment (default 0x200000) +* - a minimum address for the start of the image (default 0) +* - a maximum address for the last byte of the image (default 0xffffffff) +*/ +#define XEN_ELFNOTE_PHYS32_RELOC 19 +/* * The number of the highest elfnote defined. */ -#define XEN_ELFNOTE_MAX XEN_ELFNOTE_PHYS32_ENTRY +#define XEN_ELFNOTE_MAX XEN_ELFNOTE_PHYS32_RELOC#endif /* __XEN_PUBLIC_ELFNOTE_H__ */
Signature · 1 line
2.44.0