efi: Enable basic sandbox support for EFI loader
v6 cover efi: Enable basic sandbox support for EFI loader
A limitation of the EFI loader at present is that it does not build with sandbox. This makes it hard to write tests, since sandbox is used for most testing in U-Boot.
This series enables the EFI loader feature. It allows sandbox to build and run a trivial function which calls the EFI API to output a message.
This series is at u-boot-dm/efi-working
Changes in v6:
- Warn about building sandbox EFI support on something other than __x86_64__
Changes in v5:
- Add new patch to disallow CMD_BOOTEFI_SELFTEST on sandbox
- Drop call to efi_init_obj_list() which is now done in do_bootefi()
- Drop period after "elf" in comment
- Introduce load_options_path to specifyc U-Boot env var for load_options_path
- Rebase to master
Changes in v4:
- Move the fix to query_console_serial()
- Rebase to master
- Update SPDX tags
Changes in v3:
- Add new patch to init the 'rows' and 'cols' variables
- Add new patch to rename bootefi_test_finish() to bootefi_run_finish()
- Add new patch to split out test init/uninit into functions
- Add patch to create a function to set up for running EFI code
- Drop incorrect map_sysmem() in write_smbios_table()
- Init the 'rows' and 'cols' vars to avoid a compiler error (gcc 4.8.4)
- Rebase to master
Changes in v2:
- Rebase to master
- Update to use mapmem instead of a cast
Simon Glass (13):
efi: Don't allow CMD_BOOTEFI_SELFTEST on sandbox
efi: Init the 'rows' and 'cols' variables
efi: sandbox: Adjust memory usage for sandbox
sandbox: smbios: Update to support sandbox
efi: sandbox: Add distroboot support
efi: sandbox: Add relocation constants
efi: Add a comment about duplicated ELF constants
efi: sandbox: Enable EFI loader builder for sandbox
efi: Split out test init/uninit into functions
efi: sandbox: Add a simple 'bootefi test' command
efi: Create a function to set up for running EFI code
efi: Rename bootefi_test_finish() to bootefi_run_finish() Revert "buildman: Extract environment as part of each build"
cmd/bootefi.c | 153 ++++++++++++++++++++++---------- include/config_distro_bootcmd.h | 13 +++ include/efi_loader.h | 3 + lib/efi_loader/Kconfig | 12 ++- lib/efi_loader/Makefile | 1 + lib/efi_loader/efi_console.c | 5 +- lib/efi_loader/efi_memory.c | 31 ++++--- lib/efi_loader/efi_runtime.c | 16 ++++ lib/efi_loader/efi_test.c | 16 ++++ lib/efi_selftest/Kconfig | 2 +- lib/smbios.c | 32 +++++-- tools/buildman/builderthread.py | 10 --- tools/buildman/func_test.py | 5 -- 13 files changed, 213 insertions(+), 86 deletions(-) create mode 100644 lib/efi_loader/efi_test.c
Signature · 1 line
2.18.0.rc1.244.gcf134e6275-goog
PATCH v6 1/13 efi: Don't allow CMD_BOOTEFI_SELFTEST on sandbox
This does not work at present and gives the following error:
output: 'ld.bfd: read in flex scanner failedscripts/Makefile.lib:390: recipe for target 'lib/efi_selftest/efi_selftest_miniapp_return_efi.so' failed
It may be possible to figure this out with suitable linker magic but it does not seem to be easy. Also, we will be able to run the tests on sandbox without using the miniapp.
So for now at least, disable this option.
---
Changes in v6: None
Changes in v5:
- Add new patch to disallow CMD_BOOTEFI_SELFTEST on sandbox
Changes in v4: None
Changes in v3: None
Changes in v2: None
lib/efi_selftest/Kconfig | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
lib/efi_selftest/Kconfig +1 -1
@@ -1,6 +1,6 @@ config CMD_BOOTEFI_SELFTEST bool "Allow booting an EFI efi_selftest" - depends on CMD_BOOTEFI + depends on CMD_BOOTEFI && !SANDBOX imply FAT imply FAT_WRITE help
Signature · 1 line
2.18.0.rc1.244.gcf134e6275-goog
PATCH v6 2/13 efi: Init the 'rows' and 'cols' variables
In reply to Simon GlassThe current code causes a compiler error on gcc 4.8.4 as used by sandbox on Ubuntu 14.04, which is fairly recent. Init these variables to fix the problem.
---
Changes in v6: None
Changes in v5: None
Changes in v4:
- Move the fix to query_console_serial()
Changes in v3:
- Add new patch to init the 'rows' and 'cols' variables
Changes in v2: None
lib/efi_loader/efi_console.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-)
lib/efi_loader/efi_console.c +4 -1
@@ -204,8 +204,11 @@ static int query_console_serial(int *rows, int *cols)return -1 ;/* Read {depth,rows,cols} */ -if (term_read_reply (n,3 ,'t' )) +if (term_read_reply (n,3 ,'t' )) { + *rows =-1 ; + *cols =-1 ;return -1 ; + } *cols = n[2 ]; *rows = n[1 ];
Signature · 1 line
2.18.0.rc1.244.gcf134e6275-goog
PATCH v6 2/13 efi: Init the 'rows' and 'cols' variables
Quoted 5 lines · Simon Glass
The current code causes a compiler error on gcc 4.8.4 as used by sandbox on Ubuntu 14.04, which is fairly recent. Init these variables to fix the problem.
Signed-off-by Simon Glass <[email protected]>
I thought this patch was obsolete now?
Alex
PATCH v6 2/13 efi: Init the 'rows' and 'cols' variables
Hi Alex,
Quoted 6 lines · Alexander Graf
Quoted 5 lines · Simon Glass
The current code causes a compiler error on gcc 4.8.4 as used by sandbox on Ubuntu 14.04, which is fairly recent. Init these variables to fix the problem.
Quoted 7 lines
Quoted 1 line
Signed-off-by Simon Glass <[email protected]>I thought this patch was obsolete now?
Alex
Yes Heinrich pointed out that it isn't needed now.
Regards,
Simon
PATCH v6 3/13 efi: sandbox: Adjust memory usage for sandbox
In reply to Simon GlassWith sandbox the U-Boot code is not mapped into the sandbox memory range so does not need to be excluded when allocating EFI memory. Update the EFI memory init code to take account of that.
Also use mapmem instead of a cast to convert a memory address to a pointer.
---
Changes in v6: None
Changes in v5: None
Changes in v4: None
Changes in v3: None
Changes in v2:
- Update to use mapmem instead of a cast
lib/efi_loader/efi_memory.c | 31 ++++++++++++++++++------------- 1 file changed, 18 insertions(+), 13 deletions(-)
lib/efi_loader/efi_memory.c +18 -13
@@ -9,6 +9,7 @@#include <efi_loader.h> #include <inttypes.h> #include <malloc.h> +#include <mapmem.h> #include <watchdog.h> #include <linux/list_sort.h>
@@ -393,7 +394,7 @@ efi_status_t efi_allocate_pool(int pool_type, efi_uintn_t size, void **buffer) &t);if (r ==EFI_SUCCESS ) { -struct efi_pool_allocation *alloc = (void *)(uintptr_t )t; +struct efi_pool_allocation *alloc =map_sysmem (t, size); alloc->num_pages = num_pages; *buffer = alloc->data; }
@@ -504,18 +505,22 @@ int efi_memory_init(void)efi_add_known_memory (); -/* Add U-Boot */ - uboot_start = (gd->start_addr_sp - uboot_stack_size) & ~EFI_PAGE_MASK ; - uboot_pages = (gd->ram_top - uboot_start) >>EFI_PAGE_SHIFT ; -efi_add_memory_map (uboot_start, uboot_pages,EFI_LOADER_DATA , false); - -/* Add Runtime Services */ - runtime_start = (ulong)&__efi_runtime_start & ~EFI_PAGE_MASK ; - runtime_end = (ulong)&__efi_runtime_stop; - runtime_end = (runtime_end +EFI_PAGE_MASK ) & ~EFI_PAGE_MASK ; - runtime_pages = (runtime_end - runtime_start) >>EFI_PAGE_SHIFT ; -efi_add_memory_map (runtime_start, runtime_pages, -EFI_RUNTIME_SERVICES_CODE , false); +if (!IS_ENABLED (CONFIG_SANDBOX )) { +/* Add U-Boot */ + uboot_start = (gd->start_addr_sp - uboot_stack_size) & + ~EFI_PAGE_MASK ; + uboot_pages = (gd->ram_top - uboot_start) >>EFI_PAGE_SHIFT ; +efi_add_memory_map (uboot_start, uboot_pages,EFI_LOADER_DATA , + false); + +/* Add Runtime Services */ + runtime_start = (ulong)&__efi_runtime_start & ~EFI_PAGE_MASK ; + runtime_end = (ulong)&__efi_runtime_stop; + runtime_end = (runtime_end +EFI_PAGE_MASK ) & ~EFI_PAGE_MASK ; + runtime_pages = (runtime_end - runtime_start) >>EFI_PAGE_SHIFT ; +efi_add_memory_map (runtime_start, runtime_pages, +EFI_RUNTIME_SERVICES_CODE , false); + }#ifdef CONFIG_EFI_LOADER_BOUNCE_BUFFER /* Request a 32bit 64MB bounce buffer region */
Signature · 1 line
2.18.0.rc1.244.gcf134e6275-goog
PATCH v6 3/13 efi: sandbox: Adjust memory usage for sandbox
Quoted 33 lines · Simon Glass
With sandbox the U-Boot code is not mapped into the sandbox memory range so does not need to be excluded when allocating EFI memory. Update the EFI memory init code to take account of that.
Also use mapmem instead of a cast to convert a memory address to a pointer.
Signed-off-by Simon Glass <[email protected]>---
Changes in v6: None
Changes in v5: None
Changes in v4: None
Changes in v3: None
Changes in v2:
- Update to use mapmem instead of a castlib/efi_loader/efi_memory.c | 31 ++++++++++++++++++------------- 1 file changed, 18 insertions(+), 13 deletions(-)lib/efi_loader/efi_memory.c +1
-0@@ -9,6 +9,7 @@#include <efi_loader.h> #include <inttypes.h> #include <malloc.h> +#include <mapmem.h> #include <watchdog.h> #include <linux/list_sort.h>
lib/efi_loader/efi_memory.c efi_status_t efi_allocate_pool(int pool_type, efi_uintn_t size, void **buffer) · line 397
Quoted 5 lines
lib/efi_loader/efi_memory.c +1
-1@@ -393,7 +394,7 @@ efi_status_t efi_allocate_pool(int pool_type, efi_uintn_t size, void **buffer) &t);if (r ==EFI_SUCCESS ) { -struct efi_pool_allocation *alloc = (void *)(uintptr_t )t; +struct efi_pool_allocation *alloc =map_sysmem (t, size);
This is still the wrong spot. We don't want the conversion to happen when going from an EFI internal address to an allocation, but when determining which addresses are usable in the first place.
Quoted 16 lines
lib/efi_loader/efi_memory.c +0
-9alloc->num_pages = num_pages; *buffer = alloc->data; }@@ -504,18 +505,22 @@ int efi_memory_init(void)efi_add_known_memory (); -/* Add U-Boot */ - uboot_start = (gd->start_addr_sp - uboot_stack_size) & ~EFI_PAGE_MASK ; - uboot_pages = (gd->ram_top - uboot_start) >>EFI_PAGE_SHIFT ; -efi_add_memory_map (uboot_start, uboot_pages,EFI_LOADER_DATA , false); - -/* Add Runtime Services */ - runtime_start = (ulong)&__efi_runtime_start & ~EFI_PAGE_MASK ; - runtime_end = (ulong)&__efi_runtime_stop; - runtime_end = (runtime_end +EFI_PAGE_MASK ) & ~EFI_PAGE_MASK ;
lib/efi_loader/efi_memory.c int efi_memory_init(void) · line 514
Quoted 10 lines
lib/efi_loader/efi_memory.c +7
-3- runtime_pages = (runtime_end - runtime_start) >>EFI_PAGE_SHIFT ; -efi_add_memory_map (runtime_start, runtime_pages, -EFI_RUNTIME_SERVICES_CODE , false); +if (!IS_ENABLED (CONFIG_SANDBOX )) { +/* Add U-Boot */ + uboot_start = (gd->start_addr_sp - uboot_stack_size) & + ~EFI_PAGE_MASK ; + uboot_pages = (gd->ram_top - uboot_start) >>EFI_PAGE_SHIFT ; +efi_add_memory_map (uboot_start, uboot_pages,EFI_LOADER_DATA , + false);
Didn't we want to have these in a function?
Alex
Quoted 12 lines
lib/efi_loader/efi_memory.c +9
-0+ +/* Add Runtime Services */ + runtime_start = (ulong)&__efi_runtime_start & ~EFI_PAGE_MASK ; + runtime_end = (ulong)&__efi_runtime_stop; + runtime_end = (runtime_end +EFI_PAGE_MASK ) & ~EFI_PAGE_MASK ; + runtime_pages = (runtime_end - runtime_start) >>EFI_PAGE_SHIFT ; +efi_add_memory_map (runtime_start, runtime_pages, +EFI_RUNTIME_SERVICES_CODE , false); + }#ifdef CONFIG_EFI_LOADER_BOUNCE_BUFFER /* Request a 32bit 64MB bounce buffer region */
PATCH v6 3/13 efi: sandbox: Adjust memory usage for sandbox
Hi Alex,
Quoted 36 lines · Alexander Graf
Quoted 35 lines · Simon Glass
With sandbox the U-Boot code is not mapped into the sandbox memory range so does not need to be excluded when allocating EFI memory. Update the EFI memory init code to take account of that.
Also use mapmem instead of a cast to convert a memory address to a pointer.
Signed-off-by Simon Glass <[email protected]>---
Changes in v6: None
Changes in v5: None
Changes in v4: None
Changes in v3: None
Changes in v2:
- Update to use mapmem instead of a castlib/efi_loader/efi_memory.c | 31 ++++++++++++++++++------------- 1 file changed, 18 insertions(+), 13 deletions(-)lib/efi_loader/efi_memory.c +1
-0@@ -9,6 +9,7 @@#include <efi_loader.h> #include <inttypes.h> #include <malloc.h> +#include <mapmem.h> #include <watchdog.h> #include <linux/list_sort.h> @@ -393,7 +394,7 @@ efi_status_t efi_allocate_pool(int pool_type,
efi_uintn_tsize, void **buffer)&t);
Quoted 8 lines
lib/efi_loader/efi_memory.c efi_status_t efi_allocate_pool(int pool_type, efi_uintn_t size, void **buffer) · line 397
Quoted 3 lines
if (r == EFI_SUCCESS) {lib/efi_loader/efi_memory.c +1
-1-struct efi_pool_allocation *alloc = (void *)(uintptr_t )t; +struct efi_pool_allocation *alloc =map_sysmem (t, size);This is still the wrong spot. We don't want the conversion to happen when going from an EFI internal address to an allocation, but when determining which addresses are usable in the first place.
There seem to be two ways to do this:
1. Record addresses (ulong) in the EFI tables and use map_sysmem() before returning an address in the allocator
2. Store pointers in the EFI tables using map_sysmem(), then do no mapping in the allocator
I've gone with option 1 since:
- the EFI addresses are not pointers
- it makes sandbox consistent with other architectures which use an address rather than a pointer in EFI tables
- we don't normally do pointer arithmetic on the results of map_sysmem()
- we normally map the memory when it is used rather than when it is set up
I think you are asking for option 2. I think that would get very confusing. The addresses where things actually end up in sandbox are best kept to sandbox.
Overall I feel that you are either missing the point of sandbox addressing, or don't agree with how it is done. But it does work pretty well and we don't get a lot of confusion with sandbox pointers since we typically use the address until the last moment.
Quoted 24 lines
Quoted 24 lines
alloc->num_pages = num_pages; *buffer = alloc->data; }lib/efi_loader/efi_memory.c +4
-11@@ -504,18 +505,22 @@ int efi_memory_init(void)efi_add_known_memory (); -/* Add U-Boot */ - uboot_start = (gd->start_addr_sp - uboot_stack_size) & ~EFI_PAGE_MASK ; - uboot_pages = (gd->ram_top - uboot_start) >>EFI_PAGE_SHIFT ; -efi_add_memory_map (uboot_start, uboot_pages,EFI_LOADER_DATA , false); - -/* Add Runtime Services */ - runtime_start = (ulong)&__efi_runtime_start & ~EFI_PAGE_MASK ; - runtime_end = (ulong)&__efi_runtime_stop; - runtime_end = (runtime_end +EFI_PAGE_MASK ) & ~EFI_PAGE_MASK ; - runtime_pages = (runtime_end - runtime_start) >>EFI_PAGE_SHIFT ; -efi_add_memory_map (runtime_start, runtime_pages, -EFI_RUNTIME_SERVICES_CODE , false); +if (!IS_ENABLED (CONFIG_SANDBOX )) { +/* Add U-Boot */ + uboot_start = (gd->start_addr_sp - uboot_stack_size) & + ~EFI_PAGE_MASK ;
Quoted 8 lines
lib/efi_loader/efi_memory.c int efi_memory_init(void) · line 517
Quoted 5 lines
lib/efi_loader/efi_memory.c +3
-0+ uboot_pages = (gd->ram_top - uboot_start) >>EFI_PAGE_SHIFT ; +efi_add_memory_map (uboot_start, uboot_pages,EFI_LOADER_DATA , + false);Didn't we want to have these in a function?
Yes but I forgot. Will do for v7.
Regards,
Simon
PATCH v6 3/13 efi: sandbox: Adjust memory usage for sandbox
Quoted 60 lines · Simon Glass
Hi Alex,
Quoted 34 lines · Alexander Graf
Quoted 33 lines · Simon Glass
With sandbox the U-Boot code is not mapped into the sandbox memory range so does not need to be excluded when allocating EFI memory. Update the EFI memory init code to take account of that.
Also use mapmem instead of a cast to convert a memory address to a pointer.
Signed-off-by Simon Glass <[email protected]>---
Changes in v6: None
Changes in v5: None
Changes in v4: None
Changes in v3: None
Changes in v2:
- Update to use mapmem instead of a castlib/efi_loader/efi_memory.c | 31 ++++++++++++++++++------------- 1 file changed, 18 insertions(+), 13 deletions(-)lib/efi_loader/efi_memory.c +1
-0@@ -9,6 +9,7 @@#include <efi_loader.h> #include <inttypes.h> #include <malloc.h> +#include <mapmem.h> #include <watchdog.h> #include <linux/list_sort.h> @@ -393,7 +394,7 @@ efi_status_t efi_allocate_pool(int pool_type,
efi_uintn_tsize, void **buffer)Quoted 8 lines
lib/efi_loader/efi_memory.c efi_status_t efi_allocate_pool(int pool_type, efi_uintn_t size, void **buffer) · line 397
Quoted 4 lines
&t); if (r == EFI_SUCCESS) {lib/efi_loader/efi_memory.c +1
-1-struct efi_pool_allocation *alloc = (void *)(uintptr_t )t; +struct efi_pool_allocation *alloc =map_sysmem (t, size);This is still the wrong spot. We don't want the conversion to happen when going from an EFI internal address to an allocation, but when determining which addresses are usable in the first place.
There seem to be two ways to do this:
1. Record addresses (ulong) in the EFI tables and use
map_sysmem()before returning an address in the allocator
2. Store pointers in the EFI tables usingmap_sysmem(), then do no mapping in the allocatorI've gone with option 1 since:
- the EFI addresses are not pointers
- it makes sandbox consistent with other architectures which use an address rather than a pointer in EFI tables
- we don't normally do pointer arithmetic on the results ofmap_sysmem()
- we normally map the memory when it is used rather than when it is set up
Quoted 8 lines
I think you are asking for option 2. I think that would get very confusing. The addresses where things actually end up in sandbox are best kept to sandbox.
Overall I feel that you are either missing the point of sandbox addressing, or don't agree with how it is done. But it does work pretty well and we don't get a lot of confusion with sandbox pointers since we typically use the address until the last moment.
I've assembled a quick tree for version 2. With that I'm able to run a simple hello world efi application. Grub refuses to start because it wants memory in the low 32bit and also emits random PIO accessing functions, which obviously don't work work from user space.
But overall, I think this is the right path to tackle this:
 https://github.com/agraf/u-boot/tree/efi-sandbox
Alex
PATCH v6 3/13 efi: sandbox: Adjust memory usage for sandbox
Hi Alex,
Quoted 78 lines · Alexander Graf
Quoted 67 lines · Simon Glass
Hi Alex,
Quoted 39 lines · Alexander Graf
Quoted 37 lines · Simon Glass
With sandbox the U-Boot code is not mapped into the sandbox memory range so does not need to be excluded when allocating EFI memory. Update the EFI
memory init code to take account of that.Also use mapmem instead of a cast to convert a memory address to a pointer.
Signed-off-by Simon Glass <[email protected]>---
Changes in v6: None
Changes in v5: None
Changes in v4: None
Changes in v3: None
Changes in v2:
- Update to use mapmem instead of a castlib/efi_loader/efi_memory.c | 31 ++++++++++++++++++------------- 1 file changed, 18 insertions(+), 13 deletions(-)lib/efi_loader/efi_memory.c +1
-0@@ -9,6 +9,7 @@#include <efi_loader.h> #include <inttypes.h> #include <malloc.h> +#include <mapmem.h> #include <watchdog.h> #include <linux/list_sort.h> @@ -393,7 +394,7 @@ efi_status_t efi_allocate_pool(int pool_type,
efi_uintn_tsize, void **buffer)&t); if (r == EFI_SUCCESS) {Quoted 8 lines
Quoted 3 lines
- struct
efi_pool_allocation*alloc = (void
*)(uintptr_t)t;
+ structefi_pool_allocation*alloc =map_sysmem(t, size);This is still the wrong spot. We don't want the conversion to happen when going from an EFI internal address to an allocation, but when determining which addresses are usable in the first place.
There seem to be two ways to do this:
1. Record addresses (ulong) in the EFI tables and use
map_sysmem()before returning an address in the allocator
2. Store pointers in the EFI tables usingmap_sysmem(), then do no mapping in the allocatorI've gone with option 1 since:
- the EFI addresses are not pointers
- it makes sandbox consistent with other architectures which use an address rather than a pointer in EFI tables
- we don't normally do pointer arithmetic on the results ofmap_sysmem()
- we normally map the memory when it is used rather than when it is set upQuoted 8 lines
I think you are asking for option 2. I think that would get very confusing. The addresses where things actually end up in sandbox are best kept to sandbox.
Overall I feel that you are either missing the point of sandbox addressing, or don't agree with how it is done. But it does work pretty well and we don't get a lot of confusion with sandbox pointers since we typically use the address until the last moment.
Quoted 8 lines
I've assembled a quick tree for version 2. With that I'm able to run a simple hello world efi application. Grub refuses to start because it wants memory in the low 32bit and also emits random PIO accessing functions, which obviously don't work work from user space.
But overall, I think this is the right path to tackle this:
What do you mean by version 2? It looks like you've added one patch, so will you send that to the list?
Anyway, I hope I can convince you of the above, the way sandbox memory works.
Regards,
Simon
PATCH v6 3/13 efi: sandbox: Adjust memory usage for sandbox
Quoted 90 lines
Am 14.06.2018 um 16:12 schrieb Simon Glass <[email protected]>:
Hi Alex,
Quoted 84 lines · Alexander Graf
Quoted 68 lines · Simon Glass
Hi Alex,
Quoted 40 lines · Alexander Graf
Quoted 39 lines · Simon Glass
With sandbox the U-Boot code is not mapped into the sandbox memory range so does not need to be excluded when allocating EFI memory. Update the EFI
memory init code to take account of that.Also use mapmem instead of a cast to convert a memory address to a pointer.
Signed-off-by Simon Glass <[email protected]>---
Changes in v6: None
Changes in v5: None
Changes in v4: None
Changes in v3: None
Changes in v2:
- Update to use mapmem instead of a castlib/efi_loader/efi_memory.c | 31 ++++++++++++++++++------------- 1 file changed, 18 insertions(+), 13 deletions(-)lib/efi_loader/efi_memory.c +1
-0@@ -9,6 +9,7 @@#include <efi_loader.h> #include <inttypes.h> #include <malloc.h> +#include <mapmem.h> #include <watchdog.h> #include <linux/list_sort.h> @@ -393,7 +394,7 @@ efi_status_t efi_allocate_pool(int pool_type,
efi_uintn_tsize, void **buffer)&t); if (r == EFI_SUCCESS) {Quoted 8 lines
Quoted 3 lines
- struct
efi_pool_allocation*alloc = (void
*)(uintptr_t)t;
+ structefi_pool_allocation*alloc =map_sysmem(t, size);This is still the wrong spot. We don't want the conversion to happen when going from an EFI internal address to an allocation, but when determining which addresses are usable in the first place.
There seem to be two ways to do this:
1. Record addresses (ulong) in the EFI tables and use
map_sysmem()before returning an address in the allocator
2. Store pointers in the EFI tables usingmap_sysmem(), then do no mapping in the allocatorI've gone with option 1 since:
- the EFI addresses are not pointers
- it makes sandbox consistent with other architectures which use an address rather than a pointer in EFI tables
- we don't normally do pointer arithmetic on the results ofmap_sysmem()
- we normally map the memory when it is used rather than when it is set upQuoted 8 lines
I think you are asking for option 2. I think that would get very confusing. The addresses where things actually end up in sandbox are best kept to sandbox.
Overall I feel that you are either missing the point of sandbox addressing, or don't agree with how it is done. But it does work pretty well and we don't get a lot of confusion with sandbox pointers since we typically use the address until the last moment.
I've assembled a quick tree for version 2. With that I'm able to run a simple hello world efi application. Grub refuses to start because it wants memory in the low 32bit and also emits random PIO accessing functions, which obviously don't work work from user space.
Quoted 5 lines
Quoted 3 lines
But overall, I think this is the right path to tackle this:
What do you mean by version 2?
Option 2 is what you called it. It's the only option we have to make efi binaries work.
Quoted 2 lines
It looks like you've added one patch,
so will you send that to the list?
It's more than 1 patch. And yes, I'll send them.
Quoted 2 lines
Anyway, I hope I can convince you of the above, the way sandbox memory works.
I still dislike option 1 :)
The reason is simple: The efi memory map is available to efi payloads. It's perfectly legal for them to do a static allocation at a particular address. We also share a lot of (host) pointers for callbacks and structs already with efi applications, so there is no real point to have a split brain situation between u-boot and host pointers.
Alex
PATCH v6 3/13 efi: sandbox: Adjust memory usage for sandbox
Hi Alex,
Quoted 102 lines · Alexander Graf
Quoted 91 lines
Am 14.06.2018 um 16:12 schrieb Simon Glass <[email protected]>:
Hi Alex,
Quoted 84 lines · Alexander Graf
Quoted 68 lines · Simon Glass
Hi Alex,
Quoted 40 lines · Alexander Graf
Quoted 39 lines · Simon Glass
With sandbox the U-Boot code is not mapped into the sandbox memory range so does not need to be excluded when allocating EFI memory. Update the EFI
memory init code to take account of that.Also use mapmem instead of a cast to convert a memory address to a pointer.
Signed-off-by Simon Glass <[email protected]>---
Changes in v6: None
Changes in v5: None
Changes in v4: None
Changes in v3: None
Changes in v2:
- Update to use mapmem instead of a castlib/efi_loader/efi_memory.c | 31 ++++++++++++++++++------------- 1 file changed, 18 insertions(+), 13 deletions(-)lib/efi_loader/efi_memory.c +1
-0@@ -9,6 +9,7 @@#include <efi_loader.h> #include <inttypes.h> #include <malloc.h> +#include <mapmem.h> #include <watchdog.h> #include <linux/list_sort.h> @@ -393,7 +394,7 @@ efi_status_t efi_allocate_pool(int pool_type,
efi_uintn_tsize, void **buffer)&t); if (r == EFI_SUCCESS) {Quoted 8 lines
Quoted 3 lines
- struct
efi_pool_allocation*alloc = (void
*)(uintptr_t)t;
+ structefi_pool_allocation*alloc =map_sysmem(t, size);This is still the wrong spot. We don't want the conversion to happen when going from an EFI internal address to an allocation, but when determining which addresses are usable in the first place.
There seem to be two ways to do this:
1. Record addresses (ulong) in the EFI tables and use
map_sysmem()before returning an address in the allocator
2. Store pointers in the EFI tables usingmap_sysmem(), then do no mapping in the allocatorI've gone with option 1 since:
- the EFI addresses are not pointers
- it makes sandbox consistent with other architectures which use an address rather than a pointer in EFI tables
- we don't normally do pointer arithmetic on the results ofmap_sysmem()
- we normally map the memory when it is used rather than when it is set upQuoted 8 lines
I think you are asking for option 2. I think that would get very confusing. The addresses where things actually end up in sandbox are best kept to sandbox.
Overall I feel that you are either missing the point of sandbox addressing, or don't agree with how it is done. But it does work pretty well and we don't get a lot of confusion with sandbox pointers since we typically use the address until the last moment.
I've assembled a quick tree for version 2. With that I'm able to run a simple hello world efi application. Grub refuses to start because it wants memory in the low 32bit and also emits random PIO accessing functions, which obviously don't work work from user space.
Quoted 5 lines
Quoted 3 lines
But overall, I think this is the right path to tackle this:
What do you mean by version 2?
Option 2 is what you called it. It's the only option we have to make efi binaries work.
Quoted 2 lines
It looks like you've added one patch,
so will you send that to the list?
Quoted 8 lines
It's more than 1 patch. And yes, I'll send them.
Quoted 2 lines
Anyway, I hope I can convince you of the above, the way sandbox memory works.
I still dislike option 1 :)
The reason is simple: The efi memory map is available to efi payloads. It's perfectly legal for them to do a static allocation at a particular address. We also share a lot of (host) pointers for callbacks and structs already with efi applications, so there is no real point to have a split brain situation between u-boot and host pointers.
OK so you mean they don't have to allocate memory before using it? How then do you make sure that there is no conflict? I thought EFI was an API?
Regards,
Simon
PATCH v6 3/13 efi: sandbox: Adjust memory usage for sandbox
Quoted 114 lines
Am 14.06.2018 um 17:43 schrieb Simon Glass <[email protected]>:
Hi Alex,
Quoted 108 lines · Alexander Graf
Quoted 91 lines
Am 14.06.2018 um 16:12 schrieb Simon Glass <[email protected]>:
Hi Alex,
Quoted 84 lines
Quoted 69 lines · Alexander Graf
On 06/14/2018 02:58 PM, Simon Glass wrote:
Hi Alex,
Quoted 40 lines
Quoted 40 lines · Alexander Graf
On 06/13/2018 04:37 AM, Simon Glass wrote:
With sandbox the U-Boot code is not mapped into the sandbox memory range so does not need to be excluded when allocating EFI memory. Update the EFI
memory init code to take account of that.Also use mapmem instead of a cast to convert a memory address to a pointer.
Signed-off-by Simon Glass <[email protected]>---
Changes in v6: None
Changes in v5: None
Changes in v4: None
Changes in v3: None
Changes in v2:
- Update to use mapmem instead of a castlib/efi_loader/efi_memory.c | 31 ++++++++++++++++++------------- 1 file changed, 18 insertions(+), 13 deletions(-)lib/efi_loader/efi_memory.c +1
-0@@ -9,6 +9,7 @@#include <efi_loader.h> #include <inttypes.h> #include <malloc.h> +#include <mapmem.h> #include <watchdog.h> #include <linux/list_sort.h> @@ -393,7 +394,7 @@ efi_status_t efi_allocate_pool(int pool_type,
efi_uintn_tsize, void **buffer)&t); if (r == EFI_SUCCESS) {Quoted 8 lines
Quoted 3 lines
- struct
efi_pool_allocation*alloc = (void
*)(uintptr_t)t;
+ structefi_pool_allocation*alloc =map_sysmem(t, size);This is still the wrong spot. We don't want the conversion to happen when going from an EFI internal address to an allocation, but when determining which addresses are usable in the first place.
There seem to be two ways to do this:
1. Record addresses (ulong) in the EFI tables and use
map_sysmem()before returning an address in the allocator
2. Store pointers in the EFI tables usingmap_sysmem(), then do no mapping in the allocatorI've gone with option 1 since:
- the EFI addresses are not pointers
- it makes sandbox consistent with other architectures which use an address rather than a pointer in EFI tables
- we don't normally do pointer arithmetic on the results ofmap_sysmem()
- we normally map the memory when it is used rather than when it is set upQuoted 8 lines
I think you are asking for option 2. I think that would get very confusing. The addresses where things actually end up in sandbox are best kept to sandbox.
Overall I feel that you are either missing the point of sandbox addressing, or don't agree with how it is done. But it does work pretty well and we don't get a lot of confusion with sandbox pointers since we typically use the address until the last moment.
I've assembled a quick tree for version 2. With that I'm able to run a simple hello world efi application. Grub refuses to start because it wants memory in the low 32bit and also emits random PIO accessing functions, which obviously don't work work from user space.
Quoted 5 lines
Quoted 3 lines
But overall, I think this is the right path to tackle this:
What do you mean by version 2?
Option 2 is what you called it. It's the only option we have to make efi binaries work.
Quoted 2 lines
It looks like you've added one patch,
so will you send that to the list?It's more than 1 patch. And yes, I'll send them.
Quoted 2 lines
Anyway, I hope I can convince you of the above, the way sandbox memory works.
Quoted 7 lines
Quoted 3 lines
I still dislike option 1 :)
The reason is simple: The efi memory map is available to efi payloads. It's perfectly legal for them to do a static allocation at a particular address. We also share a lot of (host) pointers for callbacks and structs already with efi applications, so there is no real point to have a split brain situation between u-boot and host pointers.
OK so you mean they don't have to allocate memory before using it? How then do you make sure that there is no conflict? I thought EFI was an API?
You allocate it, but payloads expect that the address you pass in as address you allocate at is the pointer/address that gets returned.
In a nutshell, what you're proposing is that the efi quivalent of mmap(addr, MAP_FIXED) returns something != addr. That will break things.
Alex
Quoted 3 lines
Regards,
Simon
PATCH v6 3/13 efi: sandbox: Adjust memory usage for sandbox
Hi Alex,
Quoted 116 lines · Alexander Graf
Quoted 116 lines
Am 14.06.2018 um 17:43 schrieb Simon Glass <[email protected]>:
Hi Alex,
Quoted 110 lines · Alexander Graf
Quoted 91 lines
Am 14.06.2018 um 16:12 schrieb Simon Glass <[email protected]>:
Hi Alex,
Quoted 84 lines
Quoted 69 lines · Alexander Graf
On 06/14/2018 02:58 PM, Simon Glass wrote:
Hi Alex,
Quoted 40 lines
Quoted 40 lines · Alexander Graf
On 06/13/2018 04:37 AM, Simon Glass wrote:
With sandbox the U-Boot code is not mapped into the sandbox memory range so does not need to be excluded when allocating EFI memory. Update the EFI
memory init code to take account of that.Also use mapmem instead of a cast to convert a memory address to a pointer.
Signed-off-by Simon Glass <[email protected]>---
Changes in v6: None
Changes in v5: None
Changes in v4: None
Changes in v3: None
Changes in v2:
- Update to use mapmem instead of a castlib/efi_loader/efi_memory.c | 31 ++++++++++++++++++------------- 1 file changed, 18 insertions(+), 13 deletions(-)lib/efi_loader/efi_memory.c +1
-0@@ -9,6 +9,7 @@#include <efi_loader.h> #include <inttypes.h> #include <malloc.h> +#include <mapmem.h> #include <watchdog.h> #include <linux/list_sort.h> @@ -393,7 +394,7 @@ efi_status_t efi_allocate_pool(int pool_type,
efi_uintn_tsize, void **buffer)&t); if (r == EFI_SUCCESS) {Quoted 8 lines
Quoted 3 lines
- struct
efi_pool_allocation*alloc = (void
*)(uintptr_t)t;
+ structefi_pool_allocation*alloc =map_sysmem(t, size);This is still the wrong spot. We don't want the conversion to happen when going from an EFI internal address to an allocation, but when determining which addresses are usable in the first place.
There seem to be two ways to do this:
1. Record addresses (ulong) in the EFI tables and use
map_sysmem()before returning an address in the allocator
2. Store pointers in the EFI tables usingmap_sysmem(), then do no mapping in the allocatorI've gone with option 1 since:
- the EFI addresses are not pointers
- it makes sandbox consistent with other architectures which use an address rather than a pointer in EFI tables
- we don't normally do pointer arithmetic on the results ofmap_sysmem()
- we normally map the memory when it is used rather than when it is set upQuoted 8 lines
I think you are asking for option 2. I think that would get very confusing. The addresses where things actually end up in sandbox are best kept to sandbox.
Overall I feel that you are either missing the point of sandbox addressing, or don't agree with how it is done. But it does work pretty well and we don't get a lot of confusion with sandbox pointers since we typically use the address until the last moment.
I've assembled a quick tree for version 2. With that I'm able to run a simple hello world efi application. Grub refuses to start because it wants memory in the low 32bit and also emits random PIO accessing functions, which obviously don't work work from user space.
Quoted 5 lines
Quoted 3 lines
But overall, I think this is the right path to tackle this:
What do you mean by version 2?
Option 2 is what you called it. It's the only option we have to make efi binaries work.
Quoted 2 lines
It looks like you've added one patch,
so will you send that to the list?It's more than 1 patch. And yes, I'll send them.
Quoted 2 lines
Anyway, I hope I can convince you of the above, the way sandbox memory works.
I still dislike option 1 :)
Quoted 8 lines
Quoted 5 lines
Quoted 1 line
The reason is simple: The efi memory map is available to efi payloads. It's perfectly legal for them to do a static allocation at a particular address. We also share a lot of (host) pointers for callbacks and structs already with efi applications, so there is no real point to have a split brain situation between u-boot and host pointers.
OK so you mean they don't have to allocate memory before using it? How then do you make sure that there is no conflict? I thought EFI was an API?
You allocate it, but payloads expect that the address you pass in as address you allocate at is the pointer/address that gets returned.
Can you please point me to that? Are you referring to this call?
static efi_status_t EFIAPI efi_get_memory_map_ext(
efi_uintn_t *memory_map_size, struct efi_mem_desc *memory_map, efi_uintn_t *map_key,
efi_uintn_t *descriptor_size, uint32_t *descriptor_version)
If so, we can still support that.
Quoted 1 line
In a nutshell, what you're proposing is that the efi quivalent of
mmap(addr, MAP_FIXED)returns something != addr. That will break things.
I see this function:
efi_status_t efi_allocate_pool(int pool_type, efi_uintn_t size, void **buffer)
So there appears to be no way for people to specify the address they want. Is there another call somewhere?
NOTE: I am not proposing that the address is made available to the EFI application - that would always see the sandbox pointer.
Regards,
Simon
PATCH v6 3/13 efi: sandbox: Adjust memory usage for sandbox
Quoted 127 lines · Simon Glass
Hi Alex,
Quoted 104 lines · Alexander Graf
Quoted 104 lines
Am 14.06.2018 um 17:43 schrieb Simon Glass <[email protected]>:
Hi Alex,
Quoted 99 lines · Alexander Graf
Quoted 88 lines
Am 14.06.2018 um 16:12 schrieb Simon Glass <[email protected]>:
Hi Alex,
Quoted 81 lines
Quoted 67 lines · Alexander Graf
On 06/14/2018 02:58 PM, Simon Glass wrote:
Hi Alex,
Quoted 39 lines
Quoted 39 lines · Alexander Graf
On 06/13/2018 04:37 AM, Simon Glass wrote:
With sandbox the U-Boot code is not mapped into the sandbox memory range so does not need to be excluded when allocating EFI memory. Update the EFI
memory init code to take account of that.Also use mapmem instead of a cast to convert a memory address to a pointer.
Signed-off-by Simon Glass <[email protected]>---
Changes in v6: None
Changes in v5: None
Changes in v4: None
Changes in v3: None
Changes in v2:
- Update to use mapmem instead of a castlib/efi_loader/efi_memory.c | 31 ++++++++++++++++++------------- 1 file changed, 18 insertions(+), 13 deletions(-)lib/efi_loader/efi_memory.c +1
-0@@ -9,6 +9,7 @@#include <efi_loader.h> #include <inttypes.h> #include <malloc.h> +#include <mapmem.h> #include <watchdog.h> #include <linux/list_sort.h> @@ -393,7 +394,7 @@ efi_status_t efi_allocate_pool(int pool_type,
efi_uintn_tsize, void **buffer)&t);Quoted 8 lines
Quoted 4 lines
if (r == EFI_SUCCESS) {- struct
efi_pool_allocation*alloc = (void
*)(uintptr_t)t;
+ structefi_pool_allocation*alloc =map_sysmem(t, size);This is still the wrong spot. We don't want the conversion to happen when going from an EFI internal address to an allocation, but when determining which addresses are usable in the first place.
There seem to be two ways to do this:
1. Record addresses (ulong) in the EFI tables and use
map_sysmem()before returning an address in the allocator
2. Store pointers in the EFI tables usingmap_sysmem(), then do no mapping in the allocatorI've gone with option 1 since:
- the EFI addresses are not pointers
- it makes sandbox consistent with other architectures which use an address rather than a pointer in EFI tables
- we don't normally do pointer arithmetic on the results ofmap_sysmem()
- we normally map the memory when it is used rather than when it is set upQuoted 8 lines
I think you are asking for option 2. I think that would get very confusing. The addresses where things actually end up in sandbox are best kept to sandbox.
Overall I feel that you are either missing the point of sandbox addressing, or don't agree with how it is done. But it does work pretty well and we don't get a lot of confusion with sandbox pointers since we typically use the address until the last moment.
I've assembled a quick tree for version 2. With that I'm able to run a simple hello world efi application. Grub refuses to start because it wants memory in the low 32bit and also emits random PIO accessing functions, which obviously don't work work from user space.
Quoted 4 lines
Quoted 3 lines
But overall, I think this is the right path to tackle this:
What do you mean by version 2?
Option 2 is what you called it. It's the only option we have to make efi binaries work.
Quoted 2 lines
It looks like you've added one patch,
so will you send that to the list?It's more than 1 patch. And yes, I'll send them.
Quoted 9 lines
Quoted 7 lines
Quoted 4 lines
Quoted 1 line
Anyway, I hope I can convince you of the above, the way sandbox memory works.
I still dislike option 1 :)
The reason is simple: The efi memory map is available to efi payloads. It's perfectly legal for them to do a static allocation at a particular address. We also share a lot of (host) pointers for callbacks and structs already with efi applications, so there is no real point to have a split brain situation between u-boot and host pointers.
OK so you mean they don't have to allocate memory before using it? How then do you make sure that there is no conflict? I thought EFI was an API?
You allocate it, but payloads expect that the address you pass in as address you allocate at is the pointer/address that gets returned.
Can you please point me to that? Are you referring to this call?
static
efi_status_tEFIAPI efi_get_memory_map_ext(
efi_uintn_t*memory_map_size, structefi_mem_desc*memory_map,efi_uintn_t*map_key,
efi_uintn_t*descriptor_size,uint32_t*descriptor_version)If so, we can still support that.
Quoted 7 lines
Quoted 1 line
In a nutshell, what you're proposing is that the efi quivalent of
mmap(addr, MAP_FIXED)returns something != addr. That will break things.I see this function:
efi_status_tefi_allocate_pool(int pool_type, efi_uintn_t size, void **buffer)So there appears to be no way for people to specify the address they want. Is there another call somewhere?
You can call efi_allocate_pages() with EFI_ALLOCATE_ADDRESS or even EFI_ALLOCATE_MAX_ADDRESS which you usually base on information you receive from efi_get_memory_map().
Quoted 2 lines
NOTE: I am not proposing that the address is made available to the EFI application - that would always see the sandbox pointer.
In that case it's *much* easier to just always store host pointers in the efi memory map. Everything else will be very intrusive.
Alex
PATCH v6 3/13 efi: sandbox: Adjust memory usage for sandbox
Hi Alex,
Quoted 167 lines · Alexander Graf
Quoted 166 lines · Simon Glass
Hi Alex,
Quoted 136 lines · Alexander Graf
Quoted 136 lines
Am 14.06.2018 um 17:43 schrieb Simon Glass <[email protected]>:
Hi Alex,
Quoted 129 lines · Alexander Graf
Quoted 104 lines
Am 14.06.2018 um 16:12 schrieb Simon Glass <[email protected]>:
Hi Alex,
Quoted 97 lines
Quoted 82 lines · Alexander Graf
On 06/14/2018 02:58 PM, Simon Glass wrote:
Hi Alex,
Quoted 47 lines
Quoted 47 lines · Alexander Graf
On 06/13/2018 04:37 AM, Simon Glass wrote:
With sandbox the U-Boot code is not mapped into the sandbox memory range
so does not need to be excluded when allocating EFI memory. Update the
EFI
memory init code to take account of that.Also use mapmem instead of a cast to convert a memory address to a pointer.
Signed-off-by Simon Glass <[email protected]>---
Changes in v6: None
Changes in v5: None
Changes in v4: None
Changes in v3: None
Changes in v2:
- Update to use mapmem instead of a castlib/efi_loader/efi_memory.c | 31 ++++++++++++++++++------------- 1 file changed, 18 insertions(+), 13 deletions(-)lib/efi_loader/efi_memory.c +1
-0@@ -9,6 +9,7 @@#include <efi_loader.h> #include <inttypes.h> #include <malloc.h> +#include <mapmem.h> #include <watchdog.h> #include <linux/list_sort.h> @@ -393,7 +394,7 @@ efi_status_t efi_allocate_pool(int
pool_type,efi_uintn_tsize, void **buffer)&t); if (r == EFI_SUCCESS) {- struct
efi_pool_allocation*alloc = (void
*)(uintptr_t)t;
+ structefi_pool_allocation*alloc = map_sysmem(t,Quoted 8 lines
Quoted 1 line
size);
This is still the wrong spot. We don't want the conversion to happen when
going from an EFI internal address to an allocation, but when determining
which addresses are usable in the first place.There seem to be two ways to do this:
1. Record addresses (ulong) in the EFI tables and use
map_sysmem()before returning an address in the allocator
2. Store pointers in the EFI tables usingmap_sysmem(), then do no mapping in the allocatorI've gone with option 1 since:
- the EFI addresses are not pointers
- it makes sandbox consistent with other architectures which use an address rather than a pointer in EFI tables
- we don't normally do pointer arithmetic on the results ofmap_sysmem()
- we normally map the memory when it is used rather than when it is set upI think you are asking for option 2. I think that would get very confusing. The addresses where things actually end up in sandbox are best kept to sandbox.
Quoted 5 lines
Overall I feel that you are either missing the point of sandbox addressing, or don't agree with how it is done. But it does work pretty well and we don't get a lot of confusion with sandbox pointers
since we typically use the address until the last moment.I've assembled a quick tree for version 2. With that I'm able to run a
simple hello world efi application. Grub refuses to start because it wants
memory in the low 32bit and also emits random PIO accessing functions, which
obviously don't work work from user space.Quoted 5 lines
Quoted 3 lines
But overall, I think this is the right path to tackle this:
What do you mean by version 2?
Option 2 is what you called it. It's the only option we have to make efi binaries work.
Quoted 2 lines
It looks like you've added one patch,
so will you send that to the list?It's more than 1 patch. And yes, I'll send them.
Quoted 2 lines
Anyway, I hope I can convince you of the above, the way sandbox memory works.
I still dislike option 1 :)
The reason is simple: The efi memory map is available to efi payloads. It's perfectly legal for them to do a static allocation at a particular address. We also share a lot of (host) pointers for callbacks and structs already with efi applications, so there is no real point to have a split brain situation between u-boot and host pointers.
Quoted 7 lines
Quoted 3 lines
OK so you mean they don't have to allocate memory before using it? How then do you make sure that there is no conflict? I thought EFI was an API?
You allocate it, but payloads expect that the address you pass in as address you allocate at is the pointer/address that gets returned.
Can you please point me to that? Are you referring to this call?
static
efi_status_tEFIAPI efi_get_memory_map_ext(
efi_uintn_t*memory_map_size, structefi_mem_desc*memory_map,efi_uintn_t*map_key,
efi_uintn_t*descriptor_size,uint32_t*descriptor_version)If so, we can still support that.
Quoted 2 lines
In a nutshell, what you're proposing is that the efi quivalent of
mmap(addr, MAP_FIXED)returns something != addr. That will break things.I see this function:
efi_status_tefi_allocate_pool(int pool_type, efi_uintn_t size, void **buffer)
Quoted 7 lines
Quoted 2 lines
So there appears to be no way for people to specify the address they want. Is there another call somewhere?
You can call
efi_allocate_pages()withEFI_ALLOCATE_ADDRESSor evenEFI_ALLOCATE_MAX_ADDRESSwhich you usually base on information you receive fromefi_get_memory_map().
That's fine though, I'm not worried about that. It will work correctly, right?
Quoted 7 lines
Quoted 3 lines
NOTE: I am not proposing that the address is made available to the EFI application - that would always see the sandbox pointer.
In that case it's *much* easier to just always store host pointers in the efi memory map. Everything else will be very intrusive.
Yes there will be map_sysmem() and map_to_sysmem() calls at each boundary. That 'intrusion' is what we have everywhere in sandbox and it is how things work. But the nice thing is that the tables don't end up with 'private' sandbox addresses in them. Also see me comments above.
Regards,
Simon
PATCH v6 3/13 efi: sandbox: Adjust memory usage for sandbox
Quoted 169 lines · Simon Glass
Hi Alex,
Quoted 155 lines · Alexander Graf
Quoted 154 lines · Simon Glass
Hi Alex,
Quoted 127 lines · Alexander Graf
Quoted 127 lines
Am 14.06.2018 um 17:43 schrieb Simon Glass <[email protected]>:
Hi Alex,
Quoted 122 lines · Alexander Graf
Quoted 101 lines
Am 14.06.2018 um 16:12 schrieb Simon Glass <[email protected]>:
Hi Alex,
Quoted 94 lines
Quoted 80 lines · Alexander Graf
On 06/14/2018 02:58 PM, Simon Glass wrote:
Hi Alex,
Quoted 46 lines
Quoted 46 lines · Alexander Graf
On 06/13/2018 04:37 AM, Simon Glass wrote:
With sandbox the U-Boot code is not mapped into the sandbox memory range
so does not need to be excluded when allocating EFI memory. Update the
EFI
memory init code to take account of that.Also use mapmem instead of a cast to convert a memory address to a pointer.
Signed-off-by Simon Glass <[email protected]>---
Changes in v6: None
Changes in v5: None
Changes in v4: None
Changes in v3: None
Changes in v2:
- Update to use mapmem instead of a castlib/efi_loader/efi_memory.c | 31 ++++++++++++++++++------------- 1 file changed, 18 insertions(+), 13 deletions(-)lib/efi_loader/efi_memory.c +1
-0@@ -9,6 +9,7 @@#include <efi_loader.h> #include <inttypes.h> #include <malloc.h> +#include <mapmem.h> #include <watchdog.h> #include <linux/list_sort.h> @@ -393,7 +394,7 @@ efi_status_t efi_allocate_pool(int
pool_type,efi_uintn_tsize, void **buffer)&t); if (r == EFI_SUCCESS) {- struct
efi_pool_allocation*alloc = (void
*)(uintptr_t)t;Quoted 8 lines
Quoted 2 lines
+ struct
efi_pool_allocation*alloc = map_sysmem(t,
size);This is still the wrong spot. We don't want the conversion to happen when
going from an EFI internal address to an allocation, but when determining
which addresses are usable in the first place.There seem to be two ways to do this:
1. Record addresses (ulong) in the EFI tables and use
map_sysmem()before returning an address in the allocator
2. Store pointers in the EFI tables usingmap_sysmem(), then do no mapping in the allocatorI've gone with option 1 since:
- the EFI addresses are not pointers
- it makes sandbox consistent with other architectures which use an address rather than a pointer in EFI tables
- we don't normally do pointer arithmetic on the results ofmap_sysmem()
- we normally map the memory when it is used rather than when it is set upI think you are asking for option 2. I think that would get very confusing. The addresses where things actually end up in sandbox are best kept to sandbox.
Quoted 5 lines
Overall I feel that you are either missing the point of sandbox addressing, or don't agree with how it is done. But it does work pretty well and we don't get a lot of confusion with sandbox pointers
since we typically use the address until the last moment.I've assembled a quick tree for version 2. With that I'm able to run a
simple hello world efi application. Grub refuses to start because it wants
memory in the low 32bit and also emits random PIO accessing functions, which
obviously don't work work from user space.Quoted 4 lines
Quoted 3 lines
But overall, I think this is the right path to tackle this:
What do you mean by version 2?
Option 2 is what you called it. It's the only option we have to make efi binaries work.
Quoted 2 lines
It looks like you've added one patch,
so will you send that to the list?It's more than 1 patch. And yes, I'll send them.
Quoted 2 lines
Anyway, I hope I can convince you of the above, the way sandbox memory works.
I still dislike option 1 :)
The reason is simple: The efi memory map is available to efi payloads. It's perfectly legal for them to do a static allocation at a particular address. We also share a lot of (host) pointers for callbacks and structs already with efi applications, so there is no real point to have a split brain situation between u-boot and host pointers.
Quoted 6 lines
Quoted 3 lines
OK so you mean they don't have to allocate memory before using it? How then do you make sure that there is no conflict? I thought EFI was an API?
You allocate it, but payloads expect that the address you pass in as address you allocate at is the pointer/address that gets returned.
Can you please point me to that? Are you referring to this call?
static
efi_status_tEFIAPI efi_get_memory_map_ext(
efi_uintn_t*memory_map_size, structefi_mem_desc*memory_map,efi_uintn_t*map_key,efi_uintn_t*descriptor_size,uint32_t*descriptor_version)If so, we can still support that.
Quoted 2 lines
In a nutshell, what you're proposing is that the efi quivalent of
mmap(addr, MAP_FIXED)returns something != addr. That will break things.I see this function:
efi_status_tefi_allocate_pool(int pool_type, efi_uintn_t size, void **buffer)Quoted 6 lines
Quoted 2 lines
So there appears to be no way for people to specify the address they want. Is there another call somewhere?
You can call
efi_allocate_pages()withEFI_ALLOCATE_ADDRESSor evenEFI_ALLOCATE_MAX_ADDRESSwhich you usually base on information you receive fromefi_get_memory_map().That's fine though, I'm not worried about that. It will work correctly, right?
Quoted 3 lines
Quoted 2 lines
NOTE: I am not proposing that the address is made available to the EFI application - that would always see the sandbox pointer.
Quoted 7 lines
Quoted 2 lines
In that case it's *much* easier to just always store host pointers in the efi memory map. Everything else will be very intrusive.
Yes there will be
map_sysmem()andmap_to_sysmem()calls at each boundary. That 'intrusion' is what we have everywhere in sandbox and it is how things work. But the nice thing is that the tables don't end up with 'private' sandbox addresses in them. Also see me comments above.
I seem to be missing the point. efi_loader is just a tiny shim layer whose only purpose it is to interface between the binary efi interface and U-Boot internal code. So why shouldn't efi_loader treat addresses as pointers internally?
So how about this: I'll send out a patch set that works the way I think is the right way forward. With this I can at least run binaries generated by gnu-efi. Then you can send a patch on top with a proposal how you think the mapping should happen. My guess is that your patch will just clutter the code, but I'll be happy to be persuaded otherwise.
Alex
PATCH v6 3/13 efi: sandbox: Adjust memory usage for sandbox
Hi Alex,
Quoted 220 lines · Alexander Graf
Quoted 204 lines · Simon Glass
Hi Alex,
Quoted 184 lines · Alexander Graf
Quoted 182 lines · Simon Glass
Hi Alex,
Quoted 151 lines · Alexander Graf
Quoted 151 lines
Am 14.06.2018 um 17:43 schrieb Simon Glass <[email protected]>:
Hi Alex,
Quoted 144 lines · Alexander Graf
Quoted 114 lines
Am 14.06.2018 um 16:12 schrieb Simon Glass <[email protected]>:
Hi Alex,
Quoted 107 lines
Quoted 90 lines · Alexander Graf
On 06/14/2018 02:58 PM, Simon Glass wrote:
Hi Alex,
Quoted 52 lines
Quoted 52 lines · Alexander Graf
On 06/13/2018 04:37 AM, Simon Glass wrote:
With sandbox the U-Boot code is not mapped into the sandbox memory
range
so does not need to be excluded when allocating EFI memory. Update
the
EFI
memory init code to take account of that.Also use mapmem instead of a cast to convert a memory address to a
pointer.Signed-off-by Simon Glass <[email protected]>---
Changes in v6: None
Changes in v5: None
Changes in v4: None
Changes in v3: None
Changes in v2:
- Update to use mapmem instead of a castlib/efi_loader/efi_memory.c | 31 ++++++++++++++++++------------- 1 file changed, 18 insertions(+), 13 deletions(-)lib/efi_loader/efi_memory.c +1
-0@@ -9,6 +9,7 @@#include <efi_loader.h> #include <inttypes.h> #include <malloc.h> +#include <mapmem.h> #include <watchdog.h> #include <linux/list_sort.h> @@ -393,7 +394,7 @@ efi_status_t efi_allocate_pool(int
pool_type,efi_uintn_tsize, void **buffer)&t); if (r == EFI_SUCCESS) {- struct
efi_pool_allocation*alloc = (void
*)(uintptr_t)t;
+ structefi_pool_allocation*alloc =
map_sysmem(t,Quoted 8 lines
Quoted 1 line
size);
This is still the wrong spot. We don't want the conversion to happen when
going from an EFI internal address to an allocation, but when determining
which addresses are usable in the first place.There seem to be two ways to do this:
1. Record addresses (ulong) in the EFI tables and use
map_sysmem()before returning an address in the allocator
2. Store pointers in the EFI tables usingmap_sysmem(), then do no mapping in the allocatorI've gone with option 1 since:
- the EFI addresses are not pointers
- it makes sandbox consistent with other architectures which use an
address rather than a pointer in EFI tables
- we don't normally do pointer arithmetic on the results ofmap_sysmem()
- we normally map the memory when it is used rather than when it is
set upI think you are asking for option 2. I think that would get very confusing. The addresses where things actually end up in sandbox are
best kept to sandbox.Quoted 5 lines
Overall I feel that you are either missing the point of sandbox addressing, or don't agree with how it is done. But it does work pretty well and we don't get a lot of confusion with sandbox pointers
since we typically use the address until the last moment.I've assembled a quick tree for version 2. With that I'm able to run
a
simple hello world efi application. Grub refuses to start because it
wants
memory in the low 32bit and also emits random PIO accessing functions, which
obviously don't work work from user space.Quoted 5 lines
Quoted 3 lines
But overall, I think this is the right path to tackle this:
What do you mean by version 2?
Option 2 is what you called it. It's the only option we have to make efi binaries work.
Quoted 2 lines
It looks like you've added one patch,
so will you send that to the list?It's more than 1 patch. And yes, I'll send them.
Quoted 3 lines
Anyway, I hope I can convince you of the above, the way sandbox memory
works.I still dislike option 1 :)
The reason is simple: The efi memory map is available to efi payloads.
It's perfectly legal for them to do a static allocation at a particular
address. We also share a lot of (host) pointers for callbacks and structs
already with efi applications, so there is no real point to have a split
brain situation between u-boot and host pointers.Quoted 7 lines
Quoted 3 lines
OK so you mean they don't have to allocate memory before using it? How then do you make sure that there is no conflict? I thought EFI was an API?
You allocate it, but payloads expect that the address you pass in as address you allocate at is the pointer/address that gets returned.
Can you please point me to that? Are you referring to this call?
static
efi_status_tEFIAPI efi_get_memory_map_ext(
efi_uintn_t*memory_map_size, structefi_mem_desc*memory_map,efi_uintn_t*map_key,efi_uintn_t*descriptor_size,uint32_t*descriptor_version)If so, we can still support that.
Quoted 3 lines
In a nutshell, what you're proposing is that the efi quivalent of
mmap(addr, MAP_FIXED)returns something != addr. That will break things.I see this function:
efi_status_tefi_allocate_pool(int pool_type, efi_uintn_t size, void **buffer)Quoted 8 lines
Quoted 2 lines
So there appears to be no way for people to specify the address they want. Is there another call somewhere?
You can call
efi_allocate_pages()withEFI_ALLOCATE_ADDRESSor evenEFI_ALLOCATE_MAX_ADDRESSwhich you usually base on information you receive
fromefi_get_memory_map().That's fine though, I'm not worried about that. It will work correctly, right?
Quoted 4 lines
Quoted 2 lines
NOTE: I am not proposing that the address is made available to the EFI application - that would always see the sandbox pointer.
Quoted 8 lines
Quoted 2 lines
In that case it's *much* easier to just always store host pointers in the efi memory map. Everything else will be very intrusive.
Yes there will be
map_sysmem()andmap_to_sysmem()calls at each boundary. That 'intrusion' is what we have everywhere in sandbox and it is how things work. But the nice thing is that the tables don't end up with 'private' sandbox addresses in them. Also see me comments above.I seem to be missing the point.
efi_loaderis just a tiny shim layer whose only purpose it is to interface between the binary efi interface and U-Boot internal code. So why shouldn'tefi_loadertreat addresses as pointers internally?
Quoted 5 lines
So how about this: I'll send out a patch set that works the way I think is the right way forward. With this I can at least run binaries generated by gnu-efi. Then you can send a patch on top with a proposal how you think the mapping should happen. My guess is that your patch will just clutter the code, but I'll be happy to be persuaded otherwise.
I'm just not keen on putting sandbox pointers in internal data structures, sorry. It's not the way it is supposed to work and it makes sandbox diffferent from all the other archs. Trying to minimise the number of calls to map_sysmem() is not really the goal here.
Regards,
Simon
PATCH v6 3/13 efi: sandbox: Adjust memory usage for sandbox
Quoted 209 lines · Simon Glass
Hi Alex,
Quoted 201 lines · Alexander Graf
Quoted 187 lines · Simon Glass
Hi Alex,
Quoted 171 lines · Alexander Graf
Quoted 170 lines · Simon Glass
Hi Alex,
Quoted 142 lines · Alexander Graf
Quoted 142 lines
Am 14.06.2018 um 17:43 schrieb Simon Glass <[email protected]>:
Hi Alex,
Quoted 137 lines · Alexander Graf
Quoted 111 lines
Am 14.06.2018 um 16:12 schrieb Simon Glass <[email protected]>:
Hi Alex,
Quoted 104 lines
Quoted 88 lines · Alexander Graf
On 06/14/2018 02:58 PM, Simon Glass wrote:
Hi Alex,
Quoted 51 lines
Quoted 51 lines · Alexander Graf
On 06/13/2018 04:37 AM, Simon Glass wrote:
With sandbox the U-Boot code is not mapped into the sandbox
memory
range
so does not need to be excluded when allocating EFI memory.
Update
the
EFI
memory init code to take account of that.Also use mapmem instead of a cast to convert a memory address to
a
pointer.Signed-off-by: Simon Glass <[email protected]>
---Changes in v6: None
Changes in v5: None
Changes in v4: None
Changes in v3: None
Changes in v2:
- Update to use mapmem instead of a castlib/efi_loader/efi_memory.c | 31
++++++++++++++++++-------------
1 file changed, 18 insertions(+), 13 deletions(-)diff --git a/lib/efi_loader/efi_memory.c
b/lib/efi_loader/efi_memory.c
index ec66af98ea..210e49ee8b 100644
--- a/lib/efi_loader/efi_memory.c
+++ b/lib/efi_loader/efi_memory.c
@@ -9,6 +9,7 @@
#include <efi_loader.h>
#include <inttypes.h>
#include <malloc.h>
+#include <mapmem.h>
#include <watchdog.h>
#include <linux/list_sort.h>
@@ -393,7 +394,7 @@ efi_status_t efi_allocate_pool(int
pool_type,
efi_uintn_t size, void **buffer)
&t);
if (r == EFI_SUCCESS) {
- struct efi_pool_allocation *alloc = (void
*)(uintptr_t)t;
+ struct efi_pool_allocation *alloc =Quoted 8 lines
Quoted 2 lines
map_sysmem(t,
size);This is still the wrong spot. We don't want the conversion to happen when
going from an EFI internal address to an allocation, but when determining
which addresses are usable in the first place.There seem to be two ways to do this:
1. Record addresses (ulong) in the EFI tables and use
map_sysmem()before returning an address in the allocator
2. Store pointers in the EFI tables usingmap_sysmem(), then do no mapping in the allocatorI've gone with option 1 since:
- the EFI addresses are not pointers
- it makes sandbox consistent with other architectures which use an
address rather than a pointer in EFI tables
- we don't normally do pointer arithmetic on the results ofmap_sysmem()
- we normally map the memory when it is used rather than when it is
set upI think you are asking for option 2. I think that would get very confusing. The addresses where things actually end up in sandbox are
best kept to sandbox.Quoted 5 lines
Overall I feel that you are either missing the point of sandbox addressing, or don't agree with how it is done. But it does work pretty well and we don't get a lot of confusion with sandbox pointers
since we typically use the address until the last moment.I've assembled a quick tree for version 2. With that I'm able to run
a
simple hello world efi application. Grub refuses to start because it
wants
memory in the low 32bit and also emits random PIO accessing functions, which
obviously don't work work from user space.Quoted 4 lines
Quoted 3 lines
But overall, I think this is the right path to tackle this:
What do you mean by version 2?
Option 2 is what you called it. It's the only option we have to make efi binaries work.
Quoted 2 lines
It looks like you've added one patch,
so will you send that to the list?It's more than 1 patch. And yes, I'll send them.
Quoted 3 lines
Anyway, I hope I can convince you of the above, the way sandbox memory
works.I still dislike option 1 :)
The reason is simple: The efi memory map is available to efi payloads.
It's perfectly legal for them to do a static allocation at a particular
address. We also share a lot of (host) pointers for callbacks and structs
already with efi applications, so there is no real point to have a split
brain situation between u-boot and host pointers.Quoted 6 lines
Quoted 3 lines
OK so you mean they don't have to allocate memory before using it? How then do you make sure that there is no conflict? I thought EFI was an API?
You allocate it, but payloads expect that the address you pass in as address you allocate at is the pointer/address that gets returned.
Can you please point me to that? Are you referring to this call?
static
efi_status_tEFIAPI efi_get_memory_map_ext(
efi_uintn_t*memory_map_size, structefi_mem_desc*memory_map,efi_uintn_t*map_key,efi_uintn_t*descriptor_size,uint32_t*descriptor_version)If so, we can still support that.
Quoted 3 lines
In a nutshell, what you're proposing is that the efi quivalent of
mmap(addr, MAP_FIXED)returns something != addr. That will break things.I see this function:
efi_status_tefi_allocate_pool(int pool_type, efi_uintn_t size, void **buffer)Quoted 7 lines
Quoted 2 lines
So there appears to be no way for people to specify the address they want. Is there another call somewhere?
You can call
efi_allocate_pages()withEFI_ALLOCATE_ADDRESSor evenEFI_ALLOCATE_MAX_ADDRESSwhich you usually base on information you receive
fromefi_get_memory_map().That's fine though, I'm not worried about that. It will work correctly, right?
Quoted 3 lines
Quoted 2 lines
NOTE: I am not proposing that the address is made available to the EFI application - that would always see the sandbox pointer.
Quoted 7 lines
Quoted 2 lines
In that case it's *much* easier to just always store host pointers in the efi memory map. Everything else will be very intrusive.
Yes there will be
map_sysmem()andmap_to_sysmem()calls at each boundary. That 'intrusion' is what we have everywhere in sandbox and it is how things work. But the nice thing is that the tables don't end up with 'private' sandbox addresses in them. Also see me comments above.I seem to be missing the point.
efi_loaderis just a tiny shim layer whose only purpose it is to interface between the binary efi interface and U-Boot internal code. So why shouldn'tefi_loadertreat addresses as pointers internally?Quoted 5 lines
So how about this: I'll send out a patch set that works the way I think is the right way forward. With this I can at least run binaries generated by gnu-efi. Then you can send a patch on top with a proposal how you think the mapping should happen. My guess is that your patch will just clutter the code, but I'll be happy to be persuaded otherwise.
Quoted 4 lines
I'm just not keen on putting sandbox pointers in internal data structures, sorry. It's not the way it is supposed to work and it makes sandbox diffferent from all the other archs. Trying to minimise the number of calls to
map_sysmem()is not really the goal here.
Well, you're not the one who needs to maintain efi_loader code after every 2nd line in it ends up being some random map/unmap call ;).
But my proposal was serious. Putting virtual addresses in is trivial and gives us proper API interfaces. We can based on that still decide whether it's better to not have pointers in the map. I am not convinced yet, but open to the idea.
And yes, code maintainability is my uttermost goal of this. Things have to be simple for sandbox, otherwise they will break.
Alex
PATCH v6 3/13 efi: sandbox: Adjust memory usage for sandbox
Hi Alex,
Quoted 253 lines · Alexander Graf
Quoted 243 lines · Simon Glass
Hi Alex,
Quoted 231 lines · Alexander Graf
Quoted 215 lines · Simon Glass
Hi Alex,
Quoted 191 lines · Alexander Graf
Quoted 189 lines · Simon Glass
Hi Alex,
Quoted 156 lines · Alexander Graf
Quoted 156 lines
Am 14.06.2018 um 17:43 schrieb Simon Glass <[email protected]>:
Hi Alex,
Quoted 149 lines · Alexander Graf
Quoted 118 lines
Am 14.06.2018 um 16:12 schrieb Simon Glass <[email protected]>:
Hi Alex,
Quoted 111 lines
Quoted 94 lines · Alexander Graf
On 06/14/2018 02:58 PM, Simon Glass wrote:
Hi Alex,
Quoted 62 lines
On 14 June 2018 at 04:12, Alexander Graf <[email protected]>
wrote:On 06/13/2018 04:37 AM, Simon Glass wrote:
With sandbox the U-Boot code is not mapped into the sandbox
memory
range
so does not need to be excluded when allocating EFI memory.
Update
the
EFI
memory init code to take account of that.Also use mapmem instead of a cast to convert a memory address
to
a
pointer.Signed-off-by: Simon Glass <[email protected]>
---Changes in v6: None
Changes in v5: None
Changes in v4: None
Changes in v3: None
Changes in v2:
- Update to use mapmem instead of a castlib/efi_loader/efi_memory.c | 31
++++++++++++++++++-------------
1 file changed, 18 insertions(+), 13 deletions(-)diff --git a/lib/efi_loader/efi_memory.c
b/lib/efi_loader/efi_memory.c
index ec66af98ea..210e49ee8b 100644
--- a/lib/efi_loader/efi_memory.c
+++ b/lib/efi_loader/efi_memory.c
@@ -9,6 +9,7 @@
#include <efi_loader.h>
#include <inttypes.h>
#include <malloc.h>
+#include <mapmem.h>
#include <watchdog.h>
#include <linux/list_sort.h>
@@ -393,7 +394,7 @@ efi_status_t efi_allocate_pool(int
pool_type,
efi_uintn_t size, void **buffer)
&t);
if (r == EFI_SUCCESS) {
- struct efi_pool_allocation *alloc = (void
*)(uintptr_t)t;
+ struct efi_pool_allocation *alloc =
map_sysmem(t,
size);This is still the wrong spot. We don't want the conversion to
happen when
going from an EFI internal address to an allocation, but when
determining
which addresses are usable in the first place.There seem to be two ways to do this:
1. Record addresses (ulong) in the EFI tables and use
map_sysmem()
before returning an address in the allocator
2. Store pointers in the EFI tables usingmap_sysmem(), then do no
mapping in the allocatorI've gone with option 1 since:
- the EFI addresses are not pointers
- it makes sandbox consistent with other architectures which use an
address rather than a pointer in EFI tables
- we don't normally do pointer arithmetic on the results ofmap_sysmem()
- we normally map the memory when it is used rather than when it is
set upI think you are asking for option 2. I think that would get very confusing. The addresses where things actually end up in sandbox are
best kept to sandbox.Quoted 5 lines
Overall I feel that you are either missing the point of sandbox addressing, or don't agree with how it is done. But it does work pretty well and we don't get a lot of confusion with sandbox pointers
since we typically use the address until the last moment.I've assembled a quick tree for version 2. With that I'm able to run
a
simple hello world efi application. Grub refuses to start because it
wants
memory in the low 32bit and also emits random PIO accessing functions, which
obviously don't work work from user space.Quoted 5 lines
Quoted 3 lines
But overall, I think this is the right path to tackle this:
What do you mean by version 2?
Option 2 is what you called it. It's the only option we have to make
efi binaries work.Quoted 2 lines
It looks like you've added one patch,
so will you send that to the list?It's more than 1 patch. And yes, I'll send them.
Quoted 3 lines
Anyway, I hope I can convince you of the above, the way sandbox memory
works.I still dislike option 1 :)
The reason is simple: The efi memory map is available to efi payloads.
It's perfectly legal for them to do a static allocation at a particular
address. We also share a lot of (host) pointers for callbacks and structs
already with efi applications, so there is no real point to have a split
brain situation between u-boot and host pointers.Quoted 9 lines
Quoted 5 lines
OK so you mean they don't have to allocate memory before using it? How
then do you make sure that there is no conflict? I thought EFI was an
API?You allocate it, but payloads expect that the address you pass in as address you allocate at is the pointer/address that gets returned.
Can you please point me to that? Are you referring to this call?
static
efi_status_tEFIAPI efi_get_memory_map_ext(
efi_uintn_t*memory_map_size, structefi_mem_desc*memory_map,efi_uintn_t*map_key,efi_uintn_t*descriptor_size,uint32_t*descriptor_version)If so, we can still support that.
Quoted 3 lines
In a nutshell, what you're proposing is that the efi quivalent of
mmap(addr, MAP_FIXED)returns something != addr. That will break things.I see this function:
efi_status_tefi_allocate_pool(int pool_type, efi_uintn_t size, void **buffer)Quoted 8 lines
Quoted 2 lines
So there appears to be no way for people to specify the address they want. Is there another call somewhere?
You can call
efi_allocate_pages()withEFI_ALLOCATE_ADDRESSor evenEFI_ALLOCATE_MAX_ADDRESSwhich you usually base on information you receive
fromefi_get_memory_map().That's fine though, I'm not worried about that. It will work correctly, right?
Quoted 7 lines
Quoted 2 lines
NOTE: I am not proposing that the address is made available to the EFI application - that would always see the sandbox pointer.
In that case it's *much* easier to just always store host pointers in the
efi memory map. Everything else will be very intrusive.Quoted 5 lines
Yes there will be
map_sysmem()andmap_to_sysmem()calls at each boundary. That 'intrusion' is what we have everywhere in sandbox and it is how things work. But the nice thing is that the tables don't end up with 'private' sandbox addresses in them. Also see me comments above.I seem to be missing the point.
efi_loaderis just a tiny shim layer whose
only purpose it is to interface between the binary efi interface and U-Boot
internal code. So why shouldn'tefi_loadertreat addresses as pointers internally?Quoted 7 lines
So how about this: I'll send out a patch set that works the way I think is
the right way forward. With this I can at least run binaries generated by gnu-efi. Then you can send a patch on top with a proposal how you think the
mapping should happen. My guess is that your patch will just clutter the code, but I'll be happy to be persuaded otherwise.Quoted 4 lines
I'm just not keen on putting sandbox pointers in internal data structures, sorry. It's not the way it is supposed to work and it makes sandbox diffferent from all the other archs. Trying to minimise the number of calls to
map_sysmem()is not really the goal here.Well, you're not the one who needs to maintain
efi_loadercode after every 2nd line in it ends up being some random map/unmap call ;).
Quoted 7 lines
But my proposal was serious. Putting virtual addresses in is trivial and gives us proper API interfaces. We can based on that still decide whether it's better to not have pointers in the map. I am not convinced yet, but open to the idea.
And yes, code maintainability is my uttermost goal of this. Things have to be simple for sandbox, otherwise they will break.
A virtual address has a particular meaning. This is not what is going on with sandbox...
Making sandbox store its internal addresses in an EFI table is not a good idea. I've tried to explain why. It is not more maintainable - it just adds confusion. At present all ulong addresses are normal U-Boot addresses and pointers point into the emulated RAM. It's pretty clear for people. Please don't change that for EFI.
Can we get my patches applied and then add yours on top? I will send v7 which I think addresses all comments.
Regards,
Simon
PATCH v6 3/13 efi: sandbox: Adjust memory usage for sandbox
Quoted 240 lines · Simon Glass
Hi Alex,
Quoted 230 lines · Alexander Graf
Quoted 221 lines · Simon Glass
Hi Alex,
Quoted 211 lines · Alexander Graf
Quoted 194 lines · Simon Glass
Hi Alex,
Quoted 178 lines · Alexander Graf
Quoted 177 lines · Simon Glass
Hi Alex,
Quoted 147 lines · Alexander Graf
Quoted 147 lines
Am 14.06.2018 um 17:43 schrieb Simon Glass <[email protected]>:
Hi Alex,
Quoted 142 lines · Alexander Graf
Quoted 115 lines
Am 14.06.2018 um 16:12 schrieb Simon Glass <[email protected]>:
Hi Alex,
Quoted 108 lines
Quoted 97 lines · Alexander Graf
On 06/14/2018 02:58 PM, Simon Glass wrote:
Hi Alex,
On 14 June 2018 at 04:12, Alexander Graf <[email protected]>
wrote:On 06/13/2018 04:37 AM, Simon Glass wrote:
With sandbox the U-Boot code is not mapped into the sandbox
memory
range
so does not need to be excluded when allocating EFI memory.
Update
the
EFI
memory init code to take account of that.Also use mapmem instead of a cast to convert a memory address
to
a
pointer.Signed-off-by: Simon Glass <[email protected]>
---Changes in v6: None
Changes in v5: None
Changes in v4: None
Changes in v3: None
Changes in v2:
- Update to use mapmem instead of a castlib/efi_loader/efi_memory.c | 31
++++++++++++++++++-------------
1 file changed, 18 insertions(+), 13 deletions(-)diff --git a/lib/efi_loader/efi_memory.c
b/lib/efi_loader/efi_memory.c
index ec66af98ea..210e49ee8b 100644
--- a/lib/efi_loader/efi_memory.c
+++ b/lib/efi_loader/efi_memory.c
@@ -9,6 +9,7 @@
#include <efi_loader.h>
#include <inttypes.h>
#include <malloc.h>
+#include <mapmem.h>
#include <watchdog.h>
#include <linux/list_sort.h>
@@ -393,7 +394,7 @@ efi_status_t efi_allocate_pool(int
pool_type,
efi_uintn_t size, void **buffer)
&t);
if (r == EFI_SUCCESS) {
- struct efi_pool_allocation *alloc = (void
*)(uintptr_t)t;
+ struct efi_pool_allocation *alloc =
map_sysmem(t,
size);This is still the wrong spot. We don't want the conversion to
happen when
going from an EFI internal address to an allocation, but when
determining
which addresses are usable in the first place.
There seem to be two ways to do this:1. Record addresses (ulong) in the EFI tables and use
map_sysmem()
before returning an address in the allocator
2. Store pointers in the EFI tables using map_sysmem(), then do
no
mapping in the allocatorI've gone with option 1 since:
- the EFI addresses are not pointers
- it makes sandbox consistent with other architectures which use
an
address rather than a pointer in EFI tables
- we don't normally do pointer arithmetic on the results of
map_sysmem()
- we normally map the memory when it is used rather than when it
is
set upI think you are asking for option 2. I think that would get very
confusing. The addresses where things actually end up in sandbox
are
best kept to sandbox.Overall I feel that you are either missing the point of sandbox
addressing, or don't agree with how it is done. But it does work
pretty well and we don't get a lot of confusion with sandbox
pointers
since we typically use the address until the last moment.I've assembled a quick tree for version 2. With that I'm able to run
a
simple hello world efi application. Grub refuses to start because it
wants
memory in the low 32bit and also emits random PIO accessing functions, which
obviously don't work work from user space.Quoted 4 lines
Quoted 3 lines
But overall, I think this is the right path to tackle this:
What do you mean by version 2?
Option 2 is what you called it. It's the only option we have to make
efi binaries work.Quoted 2 lines
It looks like you've added one patch,
so will you send that to the list?It's more than 1 patch. And yes, I'll send them.
Quoted 3 lines
Anyway, I hope I can convince you of the above, the way sandbox memory
works.I still dislike option 1 :)
The reason is simple: The efi memory map is available to efi payloads.
It's perfectly legal for them to do a static allocation at a particular
address. We also share a lot of (host) pointers for callbacks and structs
already with efi applications, so there is no real point to have a split
brain situation between u-boot and host pointers.Quoted 8 lines
Quoted 5 lines
OK so you mean they don't have to allocate memory before using it? How
then do you make sure that there is no conflict? I thought EFI was an
API?You allocate it, but payloads expect that the address you pass in as address you allocate at is the pointer/address that gets returned.
Can you please point me to that? Are you referring to this call?
static
efi_status_tEFIAPI efi_get_memory_map_ext(
efi_uintn_t*memory_map_size, structefi_mem_desc*memory_map,efi_uintn_t*map_key,efi_uintn_t*descriptor_size,uint32_t*descriptor_version)If so, we can still support that.
Quoted 3 lines
In a nutshell, what you're proposing is that the efi quivalent of
mmap(addr, MAP_FIXED)returns something != addr. That will break things.I see this function:
efi_status_tefi_allocate_pool(int pool_type, efi_uintn_t size, void **buffer)Quoted 7 lines
Quoted 2 lines
So there appears to be no way for people to specify the address they want. Is there another call somewhere?
You can call
efi_allocate_pages()withEFI_ALLOCATE_ADDRESSor evenEFI_ALLOCATE_MAX_ADDRESSwhich you usually base on information you receive
fromefi_get_memory_map().That's fine though, I'm not worried about that. It will work correctly, right?
Quoted 3 lines
Quoted 2 lines
NOTE: I am not proposing that the address is made available to the EFI application - that would always see the sandbox pointer.
Quoted 8 lines
Quoted 3 lines
In that case it's *much* easier to just always store host pointers in the
efi memory map. Everything else will be very intrusive.Yes there will be
map_sysmem()andmap_to_sysmem()calls at each boundary. That 'intrusion' is what we have everywhere in sandbox and it is how things work. But the nice thing is that the tables don't end up with 'private' sandbox addresses in them. Also see me comments above.I seem to be missing the point.
efi_loaderis just a tiny shim layer whose
only purpose it is to interface between the binary efi interface and U-Boot
internal code. So why shouldn'tefi_loadertreat addresses as pointers internally?Quoted 7 lines
So how about this: I'll send out a patch set that works the way I think is
the right way forward. With this I can at least run binaries generated by gnu-efi. Then you can send a patch on top with a proposal how you think the
mapping should happen. My guess is that your patch will just clutter the code, but I'll be happy to be persuaded otherwise.Quoted 4 lines
I'm just not keen on putting sandbox pointers in internal data structures, sorry. It's not the way it is supposed to work and it makes sandbox diffferent from all the other archs. Trying to minimise the number of calls to
map_sysmem()is not really the goal here.Well, you're not the one who needs to maintain
efi_loadercode after every 2nd line in it ends up being some random map/unmap call ;).Quoted 7 lines
But my proposal was serious. Putting virtual addresses in is trivial and gives us proper API interfaces. We can based on that still decide whether it's better to not have pointers in the map. I am not convinced yet, but open to the idea.
And yes, code maintainability is my uttermost goal of this. Things have to be simple for sandbox, otherwise they will break.
Quoted 8 lines
A virtual address has a particular meaning. This is not what is going on with sandbox...
Making sandbox store its internal addresses in an EFI table is not a good idea. I've tried to explain why. It is not more maintainable - it just adds confusion. At present all ulong addresses are normal U-Boot addresses and pointers point into the emulated RAM. It's pretty clear for people. Please don't change that for EFI.
If that's your concern, let's have the efi memory map contain pointers; because that's what it really does.
Quoted 2 lines
Can we get my patches applied and then add yours on top? I will send v7 which I think addresses all comments.
The patch set I just sent contains all patches from your set that I think we need, no?
Alex
PATCH v6 3/13 efi: sandbox: Adjust memory usage for sandbox
Hi Alex,
Quoted 278 lines · Alexander Graf
Quoted 266 lines · Simon Glass
Hi Alex,
Quoted 254 lines · Alexander Graf
Quoted 243 lines · Simon Glass
Hi Alex,
Quoted 231 lines · Alexander Graf
Quoted 215 lines · Simon Glass
Hi Alex,
Quoted 191 lines · Alexander Graf
Quoted 189 lines · Simon Glass
Hi Alex,
Quoted 156 lines · Alexander Graf
Quoted 156 lines
Am 14.06.2018 um 17:43 schrieb Simon Glass <[email protected]>:
Hi Alex,
Quoted 149 lines · Alexander Graf
Quoted 118 lines
Am 14.06.2018 um 16:12 schrieb Simon Glass <[email protected]>:
Hi Alex,
Quoted 111 lines
On 14 June 2018 at 07:41, Alexander Graf <[email protected]> wrote:
On 06/14/2018 02:58 PM, Simon Glass wrote:Hi Alex,
On 14 June 2018 at 04:12, Alexander Graf <[email protected]>
wrote:On 06/13/2018 04:37 AM, Simon Glass wrote:
With sandbox the U-Boot code is not mapped into the sandbox
memory
range
so does not need to be excluded when allocating EFI memory.
Update
the
EFI
memory init code to take account of that.Also use mapmem instead of a cast to convert a memory address
to
a
pointer.Signed-off-by: Simon Glass <[email protected]>
---Changes in v6: None
Changes in v5: None
Changes in v4: None
Changes in v3: None
Changes in v2:
- Update to use mapmem instead of a castlib/efi_loader/efi_memory.c | 31
++++++++++++++++++-------------
1 file changed, 18 insertions(+), 13 deletions(-)diff --git a/lib/efi_loader/efi_memory.c
b/lib/efi_loader/efi_memory.c
index ec66af98ea..210e49ee8b 100644
--- a/lib/efi_loader/efi_memory.c
+++ b/lib/efi_loader/efi_memory.c
@@ -9,6 +9,7 @@
#include <efi_loader.h>
#include <inttypes.h>
#include <malloc.h>
+#include <mapmem.h>
#include <watchdog.h>
#include <linux/list_sort.h>
@@ -393,7 +394,7 @@ efi_status_t efi_allocate_pool(int
pool_type,
efi_uintn_t size, void **buffer)
&t);
if (r == EFI_SUCCESS) {
- struct efi_pool_allocation *alloc = (void
*)(uintptr_t)t;
+ struct efi_pool_allocation *alloc =
map_sysmem(t,
size);This is still the wrong spot. We don't want the conversion to
happen when
going from an EFI internal address to an allocation, but when
determining
which addresses are usable in the first place.There seem to be two ways to do this:
1. Record addresses (ulong) in the EFI tables and use
map_sysmem()
before returning an address in the allocator
2. Store pointers in the EFI tables using map_sysmem(), then do
no
mapping in the allocatorI've gone with option 1 since:
- the EFI addresses are not pointers
- it makes sandbox consistent with other architectures which use
an
address rather than a pointer in EFI tables
- we don't normally do pointer arithmetic on the results of
map_sysmem()
- we normally map the memory when it is used rather than when it
is
set upI think you are asking for option 2. I think that would get very
confusing. The addresses where things actually end up in sandbox
are
best kept to sandbox.Overall I feel that you are either missing the point of sandbox
addressing, or don't agree with how it is done. But it does work
pretty well and we don't get a lot of confusion with sandbox
pointers
since we typically use the address until the last moment.I've assembled a quick tree for version 2. With that I'm able to
run
a
simple hello world efi application. Grub refuses to start because
it
wants
memory in the low 32bit and also emits random PIO accessing
functions, which
obviously don't work work from user space.Quoted 5 lines
Quoted 3 lines
But overall, I think this is the right path to tackle this:
https://github.com/agraf/u-boot/tree/efi-sandbox
What do you mean by version 2?
Option 2 is what you called it. It's the only option we have to make
efi binaries work.Quoted 2 lines
It looks like you've added one patch,
so will you send that to the list?It's more than 1 patch. And yes, I'll send them.
Quoted 3 lines
Anyway, I hope I can convince you of the above, the way sandbox memory
works.I still dislike option 1 :)
The reason is simple: The efi memory map is available to efi payloads.
It's perfectly legal for them to do a static allocation at a particular
address. We also share a lot of (host) pointers for callbacks and structs
already with efi applications, so there is no real point to have a split
brain situation between u-boot and host pointers.Quoted 9 lines
Quoted 5 lines
OK so you mean they don't have to allocate memory before using it? How
then do you make sure that there is no conflict? I thought EFI was an
API?You allocate it, but payloads expect that the address you pass in as address you allocate at is the pointer/address that gets returned.
Can you please point me to that? Are you referring to this call?
static
efi_status_tEFIAPI efi_get_memory_map_ext(
efi_uintn_t*memory_map_size, structefi_mem_desc*memory_map,efi_uintn_t*map_key,efi_uintn_t*descriptor_size,uint32_t*descriptor_version)If so, we can still support that.
Quoted 3 lines
In a nutshell, what you're proposing is that the efi quivalent of
mmap(addr, MAP_FIXED)returns something != addr. That will break things.I see this function:
efi_status_tefi_allocate_pool(int pool_type, efi_uintn_t size, void **buffer)Quoted 8 lines
Quoted 2 lines
So there appears to be no way for people to specify the address they want. Is there another call somewhere?
You can call
efi_allocate_pages()withEFI_ALLOCATE_ADDRESSor evenEFI_ALLOCATE_MAX_ADDRESSwhich you usually base on information you receive
fromefi_get_memory_map().That's fine though, I'm not worried about that. It will work correctly, right?
Quoted 7 lines
Quoted 2 lines
NOTE: I am not proposing that the address is made available to the EFI application - that would always see the sandbox pointer.
In that case it's *much* easier to just always store host pointers in the
efi memory map. Everything else will be very intrusive.Quoted 5 lines
Yes there will be
map_sysmem()andmap_to_sysmem()calls at each boundary. That 'intrusion' is what we have everywhere in sandbox and it is how things work. But the nice thing is that the tables don't end up with 'private' sandbox addresses in them. Also see me comments above.I seem to be missing the point.
efi_loaderis just a tiny shim layer whose
only purpose it is to interface between the binary efi interface and U-Boot
internal code. So why shouldn'tefi_loadertreat addresses as pointers internally?Quoted 7 lines
So how about this: I'll send out a patch set that works the way I think is
the right way forward. With this I can at least run binaries generated by gnu-efi. Then you can send a patch on top with a proposal how you think the
mapping should happen. My guess is that your patch will just clutter the code, but I'll be happy to be persuaded otherwise.Quoted 4 lines
I'm just not keen on putting sandbox pointers in internal data structures, sorry. It's not the way it is supposed to work and it makes sandbox diffferent from all the other archs. Trying to minimise the number of calls to
map_sysmem()is not really the goal here.Well, you're not the one who needs to maintain
efi_loadercode after every 2nd line in it ends up being some random map/unmap call ;).Quoted 7 lines
But my proposal was serious. Putting virtual addresses in is trivial and gives us proper API interfaces. We can based on that still decide whether it's better to not have pointers in the map. I am not convinced yet, but open to the idea.
And yes, code maintainability is my uttermost goal of this. Things have to be simple for sandbox, otherwise they will break.
Quoted 8 lines
A virtual address has a particular meaning. This is not what is going on with sandbox...
Making sandbox store its internal addresses in an EFI table is not a good idea. I've tried to explain why. It is not more maintainable - it just adds confusion. At present all ulong addresses are normal U-Boot addresses and pointers point into the emulated RAM. It's pretty clear for people. Please don't change that for EFI.
Quoted 7 lines
If that's your concern, let's have the efi memory map contain pointers; because that's what it really does.
Quoted 2 lines
Can we get my patches applied and then add yours on top? I will send v7 which I think addresses all comments.
The patch set I just sent contains all patches from your set that I think we need, no?
I don't think so, you are missing several. I would like to resolve comments and get those applied and then move forward with expanding the feature set. At least that's how it normally works in U-Boot - we don't normally have the maintainer modify the patch series to his liking :-)
Regards,
Simon
PATCH v6 3/13 efi: sandbox: Adjust memory usage for sandbox
Quoted 287 lines · Simon Glass
Hi Alex,
Quoted 284 lines · Alexander Graf
Quoted 266 lines · Simon Glass
Hi Alex,
Quoted 254 lines · Alexander Graf
Quoted 243 lines · Simon Glass
Hi Alex,
Quoted 231 lines · Alexander Graf
Quoted 215 lines · Simon Glass
Hi Alex,
Quoted 191 lines · Alexander Graf
Quoted 189 lines · Simon Glass
Hi Alex,
Quoted 156 lines · Alexander Graf
Quoted 156 lines
Am 14.06.2018 um 17:43 schrieb Simon Glass <[email protected]>:
Hi Alex,
Quoted 149 lines · Alexander Graf
Quoted 123 lines
Am 14.06.2018 um 16:12 schrieb Simon Glass <[email protected]>:
Hi Alex,
On 14 June 2018 at 07:41, Alexander Graf <[email protected]> wrote:
On 06/14/2018 02:58 PM, Simon Glass wrote:Hi Alex,
On 14 June 2018 at 04:12, Alexander Graf <[email protected]>
wrote:On 06/13/2018 04:37 AM, Simon Glass wrote:
With sandbox the U-Boot code is not mapped into the sandbox
memory
range
so does not need to be excluded when allocating EFI memory.
Update
the
EFI
memory init code to take account of that.Also use mapmem instead of a cast to convert a memory address
to
a
pointer.Signed-off-by: Simon Glass <[email protected]>
---Changes in v6: None
Changes in v5: None
Changes in v4: None
Changes in v3: None
Changes in v2:
- Update to use mapmem instead of a castlib/efi_loader/efi_memory.c | 31
++++++++++++++++++-------------
1 file changed, 18 insertions(+), 13 deletions(-)diff --git a/lib/efi_loader/efi_memory.c
b/lib/efi_loader/efi_memory.c
index ec66af98ea..210e49ee8b 100644
--- a/lib/efi_loader/efi_memory.c
+++ b/lib/efi_loader/efi_memory.c
@@ -9,6 +9,7 @@
#include <efi_loader.h>
#include <inttypes.h>
#include <malloc.h>
+#include <mapmem.h>
#include <watchdog.h>
#include <linux/list_sort.h>
@@ -393,7 +394,7 @@ efi_status_t efi_allocate_pool(int
pool_type,
efi_uintn_t size, void **buffer)
&t);
if (r == EFI_SUCCESS) {
- struct efi_pool_allocation *alloc = (void
*)(uintptr_t)t;
+ struct efi_pool_allocation *alloc =
map_sysmem(t,
size);This is still the wrong spot. We don't want the conversion to
happen when
going from an EFI internal address to an allocation, but when
determining
which addresses are usable in the first place.There seem to be two ways to do this:
1. Record addresses (ulong) in the EFI tables and use
map_sysmem()
before returning an address in the allocator
2. Store pointers in the EFI tables using map_sysmem(), then do
no
mapping in the allocatorI've gone with option 1 since:
- the EFI addresses are not pointers
- it makes sandbox consistent with other architectures which use
an
address rather than a pointer in EFI tables
- we don't normally do pointer arithmetic on the results of
map_sysmem()
- we normally map the memory when it is used rather than when it
is
set upI think you are asking for option 2. I think that would get very
confusing. The addresses where things actually end up in sandbox
are
best kept to sandbox.Overall I feel that you are either missing the point of sandbox
addressing, or don't agree with how it is done. But it does work
pretty well and we don't get a lot of confusion with sandbox
pointers
since we typically use the address until the last moment.I've assembled a quick tree for version 2. With that I'm able to
run
a
simple hello world efi application. Grub refuses to start because
it
wants
memory in the low 32bit and also emits random PIO accessing
functions, which
obviously don't work work from user space.But overall, I think this is the right path to tackle this:
https://github.com/agraf/u-boot/tree/efi-sandbox
What do you mean by version 2?
Option 2 is what you called it. It's the only option we have to make
efi binaries work.Quoted 2 lines
It looks like you've added one patch,
so will you send that to the list?It's more than 1 patch. And yes, I'll send them.
Quoted 3 lines
Anyway, I hope I can convince you of the above, the way sandbox
memory
works.I still dislike option 1 :)
The reason is simple: The efi memory map is available to efi payloads.
It's perfectly legal for them to do a static allocation at a particular
address. We also share a lot of (host) pointers for callbacks and structs
already with efi applications, so there is no real point to have a split
brain situation between u-boot and host pointers.Quoted 9 lines
Quoted 5 lines
OK so you mean they don't have to allocate memory before using it? How
then do you make sure that there is no conflict? I thought EFI was an
API?You allocate it, but payloads expect that the address you pass in as address you allocate at is the pointer/address that gets returned.
Can you please point me to that? Are you referring to this call?
static
efi_status_tEFIAPI efi_get_memory_map_ext(
efi_uintn_t*memory_map_size, structefi_mem_desc*memory_map,efi_uintn_t*map_key,efi_uintn_t*descriptor_size,uint32_t*descriptor_version)If so, we can still support that.
Quoted 3 lines
In a nutshell, what you're proposing is that the efi quivalent of
mmap(addr, MAP_FIXED)returns something != addr. That will break things.I see this function:
efi_status_tefi_allocate_pool(int pool_type, efi_uintn_t size, void **buffer)Quoted 8 lines
Quoted 2 lines
So there appears to be no way for people to specify the address they want. Is there another call somewhere?
You can call
efi_allocate_pages()withEFI_ALLOCATE_ADDRESSor evenEFI_ALLOCATE_MAX_ADDRESSwhich you usually base on information you receive
fromefi_get_memory_map().That's fine though, I'm not worried about that. It will work correctly, right?
Quoted 7 lines
Quoted 2 lines
NOTE: I am not proposing that the address is made available to the EFI application - that would always see the sandbox pointer.
In that case it's *much* easier to just always store host pointers in the
efi memory map. Everything else will be very intrusive.Quoted 5 lines
Yes there will be
map_sysmem()andmap_to_sysmem()calls at each boundary. That 'intrusion' is what we have everywhere in sandbox and it is how things work. But the nice thing is that the tables don't end up with 'private' sandbox addresses in them. Also see me comments above.I seem to be missing the point.
efi_loaderis just a tiny shim layer whose
only purpose it is to interface between the binary efi interface and U-Boot
internal code. So why shouldn'tefi_loadertreat addresses as pointers internally?Quoted 7 lines
So how about this: I'll send out a patch set that works the way I think is
the right way forward. With this I can at least run binaries generated by gnu-efi. Then you can send a patch on top with a proposal how you think the
mapping should happen. My guess is that your patch will just clutter the code, but I'll be happy to be persuaded otherwise.Quoted 4 lines
I'm just not keen on putting sandbox pointers in internal data structures, sorry. It's not the way it is supposed to work and it makes sandbox diffferent from all the other archs. Trying to minimise the number of calls to
map_sysmem()is not really the goal here.Well, you're not the one who needs to maintain
efi_loadercode after every 2nd line in it ends up being some random map/unmap call ;).Quoted 7 lines
But my proposal was serious. Putting virtual addresses in is trivial and gives us proper API interfaces. We can based on that still decide whether it's better to not have pointers in the map. I am not convinced yet, but open to the idea.
And yes, code maintainability is my uttermost goal of this. Things have to be simple for sandbox, otherwise they will break.
Quoted 8 lines
A virtual address has a particular meaning. This is not what is going on with sandbox...
Making sandbox store its internal addresses in an EFI table is not a good idea. I've tried to explain why. It is not more maintainable - it just adds confusion. At present all ulong addresses are normal U-Boot addresses and pointers point into the emulated RAM. It's pretty clear for people. Please don't change that for EFI.
If that's your concern, let's have the efi memory map contain pointers; because that's what it really does.
Quoted 2 lines
Can we get my patches applied and then add yours on top? I will send v7 which I think addresses all comments.
Quoted 7 lines
Quoted 1 line
The patch set I just sent contains all patches from your set that I think we need, no?
I don't think so, you are missing several. I would like to resolve comments and get those applied and then move forward with expanding the feature set. At least that's how it normally works in U-Boot - we don't normally have the maintainer modify the patch series to his liking :-)
True, usually I would just ignore patches after my comments weren't address 5 times in a row ;).
In all seriousness, I want us to move this forward properly. So what patches from your set are missing?
[U-Boot,v7,10/10] efi: Rename bootefi_test_finish() to bootefi_run_finish() [U-Boot,v7,09/10] efi: Create a function to set up for running EFI code [U-Boot,v7,08/10] efi: sandbox: Add a simple 'bootefi test' command [U-Boot,v7,07/10] efi: Split out test init/uninit into functions
I don't want to introduce any new commands or do refactoring at this point. Let's make the selftest work first.
[U-Boot,v7,06/10] efi: sandbox: Enable EFI loader build for sandbox
Included
[U-Boot,v7,05/10] efi: sandbox: Add relocation constants
Included
[U-Boot,v7,04/10] efi: sandbox: Add distroboot support
Included
[U-Boot,v7,03/10] sandbox: smbios: Update to support sandbox
Dropped because it's no longer needed.
[U-Boot,v7,02/10] efi: sandbox: Adjust memory usage for sandbox
Dropped because it's no longer needed.
[U-Boot,v7,01/10] efi: Don't allow CMD_BOOTEFI_SELFTEST on sandbox
Replaced with Heinrich's other proposal
So I think all patches that should be in are in?
Alex
PATCH v6 3/13 efi: sandbox: Adjust memory usage for sandbox
Hi Alex,
Quoted 293 lines · Alexander Graf
Quoted 289 lines · Simon Glass
Hi Alex,
Quoted 285 lines · Alexander Graf
Quoted 266 lines · Simon Glass
Hi Alex,
Quoted 254 lines · Alexander Graf
Quoted 243 lines · Simon Glass
Hi Alex,
Quoted 231 lines · Alexander Graf
Quoted 215 lines · Simon Glass
Hi Alex,
Quoted 191 lines · Alexander Graf
Quoted 189 lines · Simon Glass
Hi Alex,
Quoted 156 lines · Alexander Graf
Quoted 156 lines
Am 14.06.2018 um 17:43 schrieb Simon Glass <[email protected]>:
Hi Alex,
Quoted 149 lines · Alexander Graf
Am 14.06.2018 um 16:12 schrieb Simon Glass <[email protected]>:
Hi Alex,
On 14 June 2018 at 07:41, Alexander Graf <[email protected]> wrote:
On 06/14/2018 02:58 PM, Simon Glass wrote:Hi Alex,
On 14 June 2018 at 04:12, Alexander Graf <[email protected]>
wrote:On 06/13/2018 04:37 AM, Simon Glass wrote:
With sandbox the U-Boot code is not mapped into the sandbox
memory
range
so does not need to be excluded when allocating EFI memory.
Update
the
EFI
memory init code to take account of that.Also use mapmem instead of a cast to convert a memory address
to
a
pointer.Signed-off-by: Simon Glass <[email protected]>
---Changes in v6: None
Changes in v5: None
Changes in v4: None
Changes in v3: None
Changes in v2:
- Update to use mapmem instead of a castlib/efi_loader/efi_memory.c | 31
++++++++++++++++++-------------
1 file changed, 18 insertions(+), 13 deletions(-)diff --git a/lib/efi_loader/efi_memory.c
b/lib/efi_loader/efi_memory.c
index ec66af98ea..210e49ee8b 100644
--- a/lib/efi_loader/efi_memory.c
+++ b/lib/efi_loader/efi_memory.c
@@ -9,6 +9,7 @@
#include <efi_loader.h>
#include <inttypes.h>
#include <malloc.h>
+#include <mapmem.h>
#include <watchdog.h>
#include <linux/list_sort.h>
@@ -393,7 +394,7 @@ efi_status_t efi_allocate_pool(int
pool_type,
efi_uintn_t size, void **buffer)
&t);
if (r == EFI_SUCCESS) {
- struct efi_pool_allocation *alloc = (void
*)(uintptr_t)t;
+ struct efi_pool_allocation *alloc =
map_sysmem(t,
size);This is still the wrong spot. We don't want the conversion to
happen when
going from an EFI internal address to an allocation, but when
determining
which addresses are usable in the first place.There seem to be two ways to do this:
1. Record addresses (ulong) in the EFI tables and use
map_sysmem()
before returning an address in the allocator
2. Store pointers in the EFI tables using map_sysmem(), then do
no
mapping in the allocatorI've gone with option 1 since:
- the EFI addresses are not pointers
- it makes sandbox consistent with other architectures which use
an
address rather than a pointer in EFI tables
- we don't normally do pointer arithmetic on the results of
map_sysmem()
- we normally map the memory when it is used rather than when it
is
set upI think you are asking for option 2. I think that would get very
confusing. The addresses where things actually end up in sandbox
are
best kept to sandbox.Overall I feel that you are either missing the point of sandbox
addressing, or don't agree with how it is done. But it does work
pretty well and we don't get a lot of confusion with sandbox
pointers
since we typically use the address until the last moment.I've assembled a quick tree for version 2. With that I'm able to
run
a
simple hello world efi application. Grub refuses to start because
it
wants
memory in the low 32bit and also emits random PIO accessing
functions, which
obviously don't work work from user space.But overall, I think this is the right path to tackle this:
https://github.com/agraf/u-boot/tree/efi-sandbox
What do you mean by version 2?
Option 2 is what you called it. It's the only option we have to
make
efi binaries work.It looks like you've added one patch,
so will you send that to the list?It's more than 1 patch. And yes, I'll send them.
Anyway, I hope I can convince you of the above, the way sandbox
memory
works.I still dislike option 1 :)
The reason is simple: The efi memory map is available to efi
payloads.
It's perfectly legal for them to do a static allocation at a
particular
address. We also share a lot of (host) pointers for callbacks and
structs
already with efi applications, so there is no real point to have a
split
brain situation between u-boot and host pointers.Quoted 9 lines
Quoted 5 lines
OK so you mean they don't have to allocate memory before using it? How
then do you make sure that there is no conflict? I thought EFI was an
API?You allocate it, but payloads expect that the address you pass in as address you allocate at is the pointer/address that gets returned.
Can you please point me to that? Are you referring to this call?
static
efi_status_tEFIAPI efi_get_memory_map_ext(
efi_uintn_t*memory_map_size, structefi_mem_desc*memory_map,efi_uintn_t*map_key,efi_uintn_t*descriptor_size,uint32_t*descriptor_version)If so, we can still support that.
Quoted 3 lines
In a nutshell, what you're proposing is that the efi quivalent of
mmap(addr, MAP_FIXED)returns something != addr. That will break things.I see this function:
efi_status_tefi_allocate_pool(int pool_type, efi_uintn_t size, void **buffer)Quoted 8 lines
Quoted 2 lines
So there appears to be no way for people to specify the address they want. Is there another call somewhere?
You can call
efi_allocate_pages()withEFI_ALLOCATE_ADDRESSor evenEFI_ALLOCATE_MAX_ADDRESSwhich you usually base on information you receive
fromefi_get_memory_map().That's fine though, I'm not worried about that. It will work correctly, right?
Quoted 7 lines
Quoted 2 lines
NOTE: I am not proposing that the address is made available to the EFI application - that would always see the sandbox pointer.
In that case it's *much* easier to just always store host pointers in the
efi memory map. Everything else will be very intrusive.Quoted 5 lines
Yes there will be
map_sysmem()andmap_to_sysmem()calls at each boundary. That 'intrusion' is what we have everywhere in sandbox and it is how things work. But the nice thing is that the tables don't end up with 'private' sandbox addresses in them. Also see me comments above.I seem to be missing the point.
efi_loaderis just a tiny shim layer whose
only purpose it is to interface between the binary efi interface and U-Boot
internal code. So why shouldn'tefi_loadertreat addresses as pointers internally?Quoted 7 lines
So how about this: I'll send out a patch set that works the way I think is
the right way forward. With this I can at least run binaries generated by gnu-efi. Then you can send a patch on top with a proposal how you think the
mapping should happen. My guess is that your patch will just clutter the code, but I'll be happy to be persuaded otherwise.Quoted 4 lines
I'm just not keen on putting sandbox pointers in internal data structures, sorry. It's not the way it is supposed to work and it makes sandbox diffferent from all the other archs. Trying to minimise the number of calls to
map_sysmem()is not really the goal here.Well, you're not the one who needs to maintain
efi_loadercode after every 2nd line in it ends up being some random map/unmap call ;).Quoted 7 lines
But my proposal was serious. Putting virtual addresses in is trivial and gives us proper API interfaces. We can based on that still decide whether it's better to not have pointers in the map. I am not convinced yet, but open to the idea.
And yes, code maintainability is my uttermost goal of this. Things have to be simple for sandbox, otherwise they will break.
Quoted 8 lines
A virtual address has a particular meaning. This is not what is going on with sandbox...
Making sandbox store its internal addresses in an EFI table is not a good idea. I've tried to explain why. It is not more maintainable - it just adds confusion. At present all ulong addresses are normal U-Boot addresses and pointers point into the emulated RAM. It's pretty clear for people. Please don't change that for EFI.
If that's your concern, let's have the efi memory map contain pointers; because that's what it really does.
Quoted 2 lines
Can we get my patches applied and then add yours on top? I will send v7 which I think addresses all comments.
The patch set I just sent contains all patches from your set that I think we need, no?
Quoted 8 lines
Quoted 5 lines
I don't think so, you are missing several. I would like to resolve comments and get those applied and then move forward with expanding the feature set. At least that's how it normally works in U-Boot - we don't normally have the maintainer modify the patch series to his liking :-)
True, usually I would just ignore patches after my comments weren't address 5 times in a row ;).
What are you talking about? I only missed one comment on v5, for one patch, which was creating the function, now done in v7. Please explain.
Quoted 21 lines
In all seriousness, I want us to move this forward properly. So what patches from your set are missing?
[U-Boot,v7,10/10] efi: Rename
bootefi_test_finish()tobootefi_run_finish()[U-Boot,v7,09/10] efi: Create a function to set up for running EFI code [U-Boot,v7,08/10] efi: sandbox: Add a simple 'bootefi test' command [U-Boot,v7,07/10] efi: Split out test init/uninit into functionsI don't want to introduce any new commands or do refactoring at this point. Let's make the selftest work first.
[U-Boot,v7,06/10] efi: sandbox: Enable EFI loader build for sandbox
Included
[U-Boot,v7,05/10] efi: sandbox: Add relocation constants
Included
Quoted 7 lines
[U-Boot,v7,04/10] efi: sandbox: Add distroboot support
Included
[U-Boot,v7,03/10] sandbox: smbios: Update to support sandbox
Dropped because it's no longer needed.
Actually this one should be applied. These are addresses so we should not be passing sandbox pointers in here.
Quoted 4 lines
[U-Boot,v7,02/10] efi: sandbox: Adjust memory usage for sandbox
Dropped because it's no longer needed.
Same here
Quoted 7 lines
[U-Boot,v7,01/10] efi: Don't allow
CMD_BOOTEFI_SELFTESTon sandboxReplaced with Heinrich's other proposal
So I think all patches that should be in are in?
I'd like to get 7-10 in as well. The 'bootefi test' thing is useful as a sanity check. Or are you saying you don't want that simple test? Otherwise you should not be changing the ordering as a maintainer.
The main issue is the sandbox memory mapping stuff, which we still disagree on.
Regards,
Simon
PATCH v6 3/13 efi: sandbox: Adjust memory usage for sandbox
Quoted 302 lines · Simon Glass
Hi Alex,
Quoted 299 lines · Alexander Graf
Quoted 294 lines · Simon Glass
Hi Alex,
Quoted 278 lines · Alexander Graf
Quoted 266 lines · Simon Glass
Hi Alex,
Quoted 254 lines · Alexander Graf
Quoted 243 lines · Simon Glass
Hi Alex,
Quoted 231 lines · Alexander Graf
Quoted 215 lines · Simon Glass
Hi Alex,
Quoted 191 lines · Alexander Graf
Quoted 189 lines · Simon Glass
Hi Alex,
Quoted 156 lines · Alexander Graf
Quoted 156 lines
Am 14.06.2018 um 17:43 schrieb Simon Glass <[email protected]>:
Hi Alex,
On 14 June 2018 at 08:22, Alexander Graf <[email protected]> wrote:
Am 14.06.2018 um 16:12 schrieb Simon Glass <[email protected]>:
Hi Alex,
On 14 June 2018 at 07:41, Alexander Graf <[email protected]> wrote:
On 06/14/2018 02:58 PM, Simon Glass wrote:Hi Alex,
On 14 June 2018 at 04:12, Alexander Graf <[email protected]>
wrote:On 06/13/2018 04:37 AM, Simon Glass wrote:
With sandbox the U-Boot code is not mapped into the sandbox
memory
range
so does not need to be excluded when allocating EFI memory.
Update
the
EFI
memory init code to take account of that.Also use mapmem instead of a cast to convert a memory address
to
a
pointer.Signed-off-by: Simon Glass <[email protected]>
---Changes in v6: None
Changes in v5: None
Changes in v4: None
Changes in v3: None
Changes in v2:
- Update to use mapmem instead of a castlib/efi_loader/efi_memory.c | 31
++++++++++++++++++-------------
1 file changed, 18 insertions(+), 13 deletions(-)diff --git a/lib/efi_loader/efi_memory.c
b/lib/efi_loader/efi_memory.c
index ec66af98ea..210e49ee8b 100644
--- a/lib/efi_loader/efi_memory.c
+++ b/lib/efi_loader/efi_memory.c
@@ -9,6 +9,7 @@
#include <efi_loader.h>
#include <inttypes.h>
#include <malloc.h>
+#include <mapmem.h>
#include <watchdog.h>
#include <linux/list_sort.h>
@@ -393,7 +394,7 @@ efi_status_t efi_allocate_pool(int
pool_type,
efi_uintn_t size, void **buffer)
&t);
if (r == EFI_SUCCESS) {
- struct efi_pool_allocation *alloc = (void
*)(uintptr_t)t;
+ struct efi_pool_allocation *alloc =
map_sysmem(t,
size);This is still the wrong spot. We don't want the conversion to
happen when
going from an EFI internal address to an allocation, but when
determining
which addresses are usable in the first place.There seem to be two ways to do this:
1. Record addresses (ulong) in the EFI tables and use
map_sysmem()
before returning an address in the allocator
2. Store pointers in the EFI tables using map_sysmem(), then do
no
mapping in the allocatorI've gone with option 1 since:
- the EFI addresses are not pointers
- it makes sandbox consistent with other architectures which use
an
address rather than a pointer in EFI tables
- we don't normally do pointer arithmetic on the results of
map_sysmem()
- we normally map the memory when it is used rather than when it
is
set upI think you are asking for option 2. I think that would get very
confusing. The addresses where things actually end up in sandbox
are
best kept to sandbox.Overall I feel that you are either missing the point of sandbox
addressing, or don't agree with how it is done. But it does work
pretty well and we don't get a lot of confusion with sandbox
pointers
since we typically use the address until the last moment.I've assembled a quick tree for version 2. With that I'm able to
run
a
simple hello world efi application. Grub refuses to start because
it
wants
memory in the low 32bit and also emits random PIO accessing
functions, which
obviously don't work work from user space.But overall, I think this is the right path to tackle this:
https://github.com/agraf/u-boot/tree/efi-sandbox
What do you mean by version 2?
Option 2 is what you called it. It's the only option we have to
make
efi binaries work.It looks like you've added one patch,
so will you send that to the list?It's more than 1 patch. And yes, I'll send them.
Anyway, I hope I can convince you of the above, the way sandbox
memory
works.I still dislike option 1 :)
The reason is simple: The efi memory map is available to efi
payloads.
It's perfectly legal for them to do a static allocation at a
particular
address. We also share a lot of (host) pointers for callbacks and
structs
already with efi applications, so there is no real point to have a
split
brain situation between u-boot and host pointers.Quoted 9 lines
Quoted 5 lines
OK so you mean they don't have to allocate memory before using it?
How
then do you make sure that there is no conflict? I thought EFI was
an
API?You allocate it, but payloads expect that the address you pass in as address you allocate at is the pointer/address that gets returned.
Can you please point me to that? Are you referring to this call?
static
efi_status_tEFIAPI efi_get_memory_map_ext(
efi_uintn_t*memory_map_size, structefi_mem_desc*memory_map,efi_uintn_t*map_key,efi_uintn_t*descriptor_size,uint32_t*descriptor_version)If so, we can still support that.
Quoted 3 lines
In a nutshell, what you're proposing is that the efi quivalent of
mmap(addr, MAP_FIXED)returns something != addr. That will break things.I see this function:
efi_status_tefi_allocate_pool(int pool_type, efi_uintn_t size, void **buffer)Quoted 8 lines
Quoted 2 lines
So there appears to be no way for people to specify the address they want. Is there another call somewhere?
You can call
efi_allocate_pages()withEFI_ALLOCATE_ADDRESSor evenEFI_ALLOCATE_MAX_ADDRESSwhich you usually base on information you receive
fromefi_get_memory_map().That's fine though, I'm not worried about that. It will work correctly, right?
Quoted 7 lines
Quoted 2 lines
NOTE: I am not proposing that the address is made available to the EFI application - that would always see the sandbox pointer.
In that case it's *much* easier to just always store host pointers in the
efi memory map. Everything else will be very intrusive.Quoted 5 lines
Yes there will be
map_sysmem()andmap_to_sysmem()calls at each boundary. That 'intrusion' is what we have everywhere in sandbox and it is how things work. But the nice thing is that the tables don't end up with 'private' sandbox addresses in them. Also see me comments above.I seem to be missing the point.
efi_loaderis just a tiny shim layer whose
only purpose it is to interface between the binary efi interface and U-Boot
internal code. So why shouldn'tefi_loadertreat addresses as pointers internally?Quoted 7 lines
So how about this: I'll send out a patch set that works the way I think is
the right way forward. With this I can at least run binaries generated by gnu-efi. Then you can send a patch on top with a proposal how you think the
mapping should happen. My guess is that your patch will just clutter the code, but I'll be happy to be persuaded otherwise.Quoted 4 lines
I'm just not keen on putting sandbox pointers in internal data structures, sorry. It's not the way it is supposed to work and it makes sandbox diffferent from all the other archs. Trying to minimise the number of calls to
map_sysmem()is not really the goal here.Well, you're not the one who needs to maintain
efi_loadercode after every 2nd line in it ends up being some random map/unmap call ;).Quoted 7 lines
But my proposal was serious. Putting virtual addresses in is trivial and gives us proper API interfaces. We can based on that still decide whether it's better to not have pointers in the map. I am not convinced yet, but open to the idea.
And yes, code maintainability is my uttermost goal of this. Things have to be simple for sandbox, otherwise they will break.
Quoted 8 lines
A virtual address has a particular meaning. This is not what is going on with sandbox...
Making sandbox store its internal addresses in an EFI table is not a good idea. I've tried to explain why. It is not more maintainable - it just adds confusion. At present all ulong addresses are normal U-Boot addresses and pointers point into the emulated RAM. It's pretty clear for people. Please don't change that for EFI.
Quoted 7 lines
If that's your concern, let's have the efi memory map contain pointers; because that's what it really does.
Quoted 2 lines
Can we get my patches applied and then add yours on top? I will send v7 which I think addresses all comments.
The patch set I just sent contains all patches from your set that I think we need, no?
I don't think so, you are missing several. I would like to resolve comments and get those applied and then move forward with expanding the feature set. At least that's how it normally works in U-Boot - we don't normally have the maintainer modify the patch series to his liking :-)
Quoted 6 lines
Quoted 2 lines
True, usually I would just ignore patches after my comments weren't address 5 times in a row ;).
What are you talking about? I only missed one comment on v5, for one patch, which was creating the function, now done in v7. Please explain.
I thought I had expressed my dislike for the current memory mapping model a few times by now. Maybe I wasn't explicit enough?
Quoted 22 lines
Quoted 22 lines
In all seriousness, I want us to move this forward properly. So what patches from your set are missing?
[U-Boot,v7,10/10] efi: Rename
bootefi_test_finish()tobootefi_run_finish()[U-Boot,v7,09/10] efi: Create a function to set up for running EFI code [U-Boot,v7,08/10] efi: sandbox: Add a simple 'bootefi test' command [U-Boot,v7,07/10] efi: Split out test init/uninit into functionsI don't want to introduce any new commands or do refactoring at this point. Let's make the selftest work first.
[U-Boot,v7,06/10] efi: sandbox: Enable EFI loader build for sandbox
Included
[U-Boot,v7,05/10] efi: sandbox: Add relocation constants
Included
[U-Boot,v7,04/10] efi: sandbox: Add distroboot support
Quoted 8 lines
Quoted 5 lines
Included
[U-Boot,v7,03/10] sandbox: smbios: Update to support sandbox
Dropped because it's no longer needed.
Actually this one should be applied. These are addresses so we should not be passing sandbox pointers in here.
Well, maybe. But then we'd have to adjust for the API the other way around, as now pointers are what allocations return (because they're a payload visible API, so they have to).
Quoted 7 lines
Quoted 5 lines
[U-Boot,v7,02/10] efi: sandbox: Adjust memory usage for sandbox
Dropped because it's no longer needed.
Same here
Why does it hurt to expose the U-Boot and RTS memory addresses?
Quoted 11 lines
Quoted 8 lines
[U-Boot,v7,01/10] efi: Don't allow
CMD_BOOTEFI_SELFTESTon sandboxReplaced with Heinrich's other proposal
So I think all patches that should be in are in?
I'd like to get 7-10 in as well. The 'bootefi test' thing is useful as a sanity check. Or are you saying you don't want that simple test?
I want a simple test, but I think the simple test is bootefi selftest. You can limit its scope to only emit "hello world" by setting an environment variable if you want to :).
Quoted 3 lines
Otherwise you should not be changing the ordering as a maintainer.
The main issue is the sandbox memory mapping stuff, which we still disagree on.
I guess we agree that we disagree :).
Looking at the current code base, we pretty much only call the map/unmap dance every time we go from command line arguments (where you pass in u-boot addresses) to pointers. Anything internal is usually dealing with pointers.
I think that's a reasonable thing to do. And I think we should follow the same logic in the efi_loader parts. Command line parameters can easily be handled via map/unmap, but anything internal - and the efi memory map is what I consider internal - should just contain pointers.
The only reason they're not marked as pointers in the struct is that we want to copy the structs right back into the payload and there they are ABI defined as 64bit fields.
Alex
PATCH v6 3/13 efi: sandbox: Adjust memory usage for sandbox
Hi Alex,
Quoted 308 lines · Alexander Graf
Quoted 305 lines · Simon Glass
Hi Alex,
Quoted 301 lines · Alexander Graf
Quoted 287 lines · Simon Glass
Hi Alex,
Quoted 284 lines · Alexander Graf
Quoted 266 lines · Simon Glass
Hi Alex,
Quoted 254 lines · Alexander Graf
Quoted 243 lines · Simon Glass
Hi Alex,
Quoted 231 lines · Alexander Graf
Quoted 215 lines · Simon Glass
Hi Alex,
Quoted 191 lines · Alexander Graf
Quoted 189 lines · Simon Glass
Hi Alex,
Quoted 165 lines · Alexander Graf
Am 14.06.2018 um 17:43 schrieb Simon Glass <[email protected]>:
Hi Alex,
On 14 June 2018 at 08:22, Alexander Graf <[email protected]> wrote:
Am 14.06.2018 um 16:12 schrieb Simon Glass <[email protected]>:
Hi Alex,
On 14 June 2018 at 07:41, Alexander Graf <[email protected]> wrote:
On 06/14/2018 02:58 PM, Simon Glass wrote:Hi Alex,
On 14 June 2018 at 04:12, Alexander Graf <[email protected]>
wrote:On 06/13/2018 04:37 AM, Simon Glass wrote:
With sandbox the U-Boot code is not mapped into the sandbox
memory
range
so does not need to be excluded when allocating EFI memory.
Update
the
EFI
memory init code to take account of that.Also use mapmem instead of a cast to convert a memory address
to
a
pointer.Signed-off-by: Simon Glass <[email protected]>
---Changes in v6: None
Changes in v5: None
Changes in v4: None
Changes in v3: None
Changes in v2:
- Update to use mapmem instead of a castlib/efi_loader/efi_memory.c | 31
++++++++++++++++++-------------
1 file changed, 18 insertions(+), 13 deletions(-)diff --git a/lib/efi_loader/efi_memory.c
b/lib/efi_loader/efi_memory.c
index ec66af98ea..210e49ee8b 100644
--- a/lib/efi_loader/efi_memory.c
+++ b/lib/efi_loader/efi_memory.c
@@ -9,6 +9,7 @@
#include <efi_loader.h>
#include <inttypes.h>
#include <malloc.h>
+#include <mapmem.h>
#include <watchdog.h>
#include <linux/list_sort.h>
@@ -393,7 +394,7 @@ efi_status_t efi_allocate_pool(int
pool_type,
efi_uintn_t size, void **buffer)
&t);
if (r == EFI_SUCCESS) {
- struct efi_pool_allocation *alloc = (void
*)(uintptr_t)t;
+ struct efi_pool_allocation *alloc =
map_sysmem(t,
size);This is still the wrong spot. We don't want the conversion to
happen when
going from an EFI internal address to an allocation, but when
determining
which addresses are usable in the first place.There seem to be two ways to do this:
1. Record addresses (ulong) in the EFI tables and use
map_sysmem()
before returning an address in the allocator
2. Store pointers in the EFI tables using map_sysmem(), then do
no
mapping in the allocatorI've gone with option 1 since:
- the EFI addresses are not pointers
- it makes sandbox consistent with other architectures which use
an
address rather than a pointer in EFI tables
- we don't normally do pointer arithmetic on the results of
map_sysmem()
- we normally map the memory when it is used rather than when it
is
set upI think you are asking for option 2. I think that would get very
confusing. The addresses where things actually end up in sandbox
are
best kept to sandbox.Overall I feel that you are either missing the point of sandbox
addressing, or don't agree with how it is done. But it does work
pretty well and we don't get a lot of confusion with sandbox
pointers
since we typically use the address until the last moment.I've assembled a quick tree for version 2. With that I'm able to
run
a
simple hello world efi application. Grub refuses to start because
it
wants
memory in the low 32bit and also emits random PIO accessing
functions, which
obviously don't work work from user space.But overall, I think this is the right path to tackle this:
https://github.com/agraf/u-boot/tree/efi-sandbox
What do you mean by version 2?
Option 2 is what you called it. It's the only option we have to
make
efi binaries work.It looks like you've added one patch,
so will you send that to the list?It's more than 1 patch. And yes, I'll send them.
Anyway, I hope I can convince you of the above, the way sandbox
memory
works.I still dislike option 1 :)
The reason is simple: The efi memory map is available to efi
payloads.
It's perfectly legal for them to do a static allocation at a
particular
address. We also share a lot of (host) pointers for callbacks and
structs
already with efi applications, so there is no real point to have a
split
brain situation between u-boot and host pointers.OK so you mean they don't have to allocate memory before using it?
How
then do you make sure that there is no conflict? I thought EFI was
an
API?You allocate it, but payloads expect that the address you pass in as
address you allocate at is the pointer/address that gets returned.Can you please point me to that? Are you referring to this call?
static
efi_status_tEFIAPI efi_get_memory_map_ext(
efi_uintn_t*memory_map_size, structefi_mem_desc*memory_map,efi_uintn_t*map_key,efi_uintn_t*descriptor_size,uint32_t*descriptor_version)If so, we can still support that.
Quoted 3 lines
In a nutshell, what you're proposing is that the efi quivalent of
mmap(addr, MAP_FIXED) returns something != addr. That will break
things.I see this function:
efi_status_tefi_allocate_pool(int pool_type, efi_uintn_t size, void **buffer)Quoted 8 lines
Quoted 2 lines
So there appears to be no way for people to specify the address they want. Is there another call somewhere?
You can call
efi_allocate_pages()withEFI_ALLOCATE_ADDRESSor evenEFI_ALLOCATE_MAX_ADDRESSwhich you usually base on information you receive
fromefi_get_memory_map().That's fine though, I'm not worried about that. It will work correctly, right?
Quoted 7 lines
Quoted 2 lines
NOTE: I am not proposing that the address is made available to the EFI application - that would always see the sandbox pointer.
In that case it's *much* easier to just always store host pointers in the
efi memory map. Everything else will be very intrusive.Quoted 5 lines
Yes there will be
map_sysmem()andmap_to_sysmem()calls at each boundary. That 'intrusion' is what we have everywhere in sandbox and it is how things work. But the nice thing is that the tables don't end up with 'private' sandbox addresses in them. Also see me comments above.I seem to be missing the point.
efi_loaderis just a tiny shim layer whose
only purpose it is to interface between the binary efi interface and U-Boot
internal code. So why shouldn'tefi_loadertreat addresses as pointers internally?Quoted 7 lines
So how about this: I'll send out a patch set that works the way I think is
the right way forward. With this I can at least run binaries generated by gnu-efi. Then you can send a patch on top with a proposal how you think the
mapping should happen. My guess is that your patch will just clutter the code, but I'll be happy to be persuaded otherwise.Quoted 4 lines
I'm just not keen on putting sandbox pointers in internal data structures, sorry. It's not the way it is supposed to work and it makes sandbox diffferent from all the other archs. Trying to minimise the number of calls to
map_sysmem()is not really the goal here.Well, you're not the one who needs to maintain
efi_loadercode after every 2nd line in it ends up being some random map/unmap call ;).Quoted 7 lines
But my proposal was serious. Putting virtual addresses in is trivial and gives us proper API interfaces. We can based on that still decide whether it's better to not have pointers in the map. I am not convinced yet, but open to the idea.
And yes, code maintainability is my uttermost goal of this. Things have to be simple for sandbox, otherwise they will break.
Quoted 8 lines
A virtual address has a particular meaning. This is not what is going on with sandbox...
Making sandbox store its internal addresses in an EFI table is not a good idea. I've tried to explain why. It is not more maintainable - it just adds confusion. At present all ulong addresses are normal U-Boot addresses and pointers point into the emulated RAM. It's pretty clear for people. Please don't change that for EFI.
If that's your concern, let's have the efi memory map contain pointers; because that's what it really does.
Quoted 2 lines
Can we get my patches applied and then add yours on top? I will send v7 which I think addresses all comments.
Quoted 7 lines
Quoted 1 line
The patch set I just sent contains all patches from your set that I think we need, no?
I don't think so, you are missing several. I would like to resolve comments and get those applied and then move forward with expanding the feature set. At least that's how it normally works in U-Boot - we don't normally have the maintainer modify the patch series to his liking :-)
True, usually I would just ignore patches after my comments weren't address 5 times in a row ;).
Quoted 6 lines
Quoted 3 lines
What are you talking about? I only missed one comment on v5, for one patch, which was creating the function, now done in v7. Please explain.
I thought I had expressed my dislike for the current memory mapping model a few times by now. Maybe I wasn't explicit enough?
I was pretty clear that I don't want to change the sandbox memory model, so that's why I didn't do anything on that point. I don't think I could have been any clearer.
Quoted 27 lines
Quoted 27 lines
Quoted 27 lines
In all seriousness, I want us to move this forward properly. So what patches from your set are missing?
[U-Boot,v7,10/10] efi: Rename
bootefi_test_finish()tobootefi_run_finish()[U-Boot,v7,09/10] efi: Create a function to set up for running EFI code [U-Boot,v7,08/10] efi: sandbox: Add a simple 'bootefi test' command [U-Boot,v7,07/10] efi: Split out test init/uninit into functionsI don't want to introduce any new commands or do refactoring at this point. Let's make the selftest work first.
[U-Boot,v7,06/10] efi: sandbox: Enable EFI loader build for sandbox
Included
[U-Boot,v7,05/10] efi: sandbox: Add relocation constants
Included
[U-Boot,v7,04/10] efi: sandbox: Add distroboot support
Included
[U-Boot,v7,03/10] sandbox: smbios: Update to support sandbox
Quoted 8 lines
Quoted 4 lines
Quoted 1 line
Dropped because it's no longer needed.
Actually this one should be applied. These are addresses so we should not be passing sandbox pointers in here.
Well, maybe. But then we'd have to adjust for the API the other way around, as now pointers are what allocations return (because they're a payload visible API, so they have to).
Actually the function I am talking about is internal to U-Boot. The _ext() function can do the mapping as needed, as per my latest series. But I feel it can still be improved, since addresses and pointers are still a little mixed together in the EFI code.
Quoted 10 lines
Quoted 8 lines
Quoted 6 lines
[U-Boot,v7,02/10] efi: sandbox: Adjust memory usage for sandbox
Dropped because it's no longer needed.
Same here
Why does it hurt to expose the U-Boot and RTS memory addresses?
What is an RTS memory address?
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
[U-Boot,v7,01/10] efi: Don't allow
CMD_BOOTEFI_SELFTESTon sandboxReplaced with Heinrich's other proposal
Quoted 8 lines
Quoted 4 lines
Quoted 1 line
So I think all patches that should be in are in?
I'd like to get 7-10 in as well. The 'bootefi test' thing is useful as a sanity check. Or are you saying you don't want that simple test?
I want a simple test, but I think the simple test is bootefi selftest. You can limit its scope to only emit "hello world" by setting an environment variable if you want to :).
I've reviewed selftest as it was built up and again recently. It is certainly not a simple sanity chec, which is what hello word is for.
Quoted 11 lines
Quoted 4 lines
Otherwise you should not be changing the ordering as a maintainer.
The main issue is the sandbox memory mapping stuff, which we still disagree on.
I guess we agree that we disagree :).
Looking at the current code base, we pretty much only call the map/unmap dance every time we go from command line arguments (where you pass in u-boot addresses) to pointers. Anything internal is usually dealing with pointers.
That's just not the case, as I have explained. Images in U-Boot, device tree, frame buffer, all peripheral's register locations, bootm, FIT and many other things are dealt with with addresses. U-Boot uses addresses for addresses, not pointers. The pointers are created from the addresses, but are not the primary means for communicating addresses.
Quoted 9 lines
I think that's a reasonable thing to do. And I think we should follow the same logic in the
efi_loaderparts. Command line parameters can easily be handled via map/unmap, but anything internal - and the efi memory map is what I consider internal - should just contain pointers.The only reason they're not marked as pointers in the struct is that we want to copy the structs right back into the payload and there they are ABI defined as 64bit fields.
I still strongly disagree with this. I would like to get the sandbox feature in without changing how sandbox works.
Regards,
Simon
PATCH v6 4/13 sandbox: smbios: Update to support sandbox
In reply to Simon GlassAt present this code casts addresses to pointers so cannot be used with sandbox. Update it to use mapmem instead.
---
Changes in v6: None
Changes in v5: None
Changes in v4: None
Changes in v3:
- Drop incorrect map_sysmem() in write_smbios_table()
Changes in v2: None
lib/smbios.c | 32 ++++++++++++++++++++++++-------- 1 file changed, 24 insertions(+), 8 deletions(-)
lib/smbios.c +24 -8
@@ -6,6 +6,7 @@*/ #include <common.h> +#include <mapmem.h> #include <smbios.h> #include <tables_csum.h> #include <version.h>
@@ -72,9 +73,10 @@ static int smbios_string_table_len(char *start)static int smbios_write_type0 (ulong *current,int handle) { -struct smbios_type0 *t = (struct smbios_type0 *)*current; +struct smbios_type0 *t;int len =sizeof (struct smbios_type0 ); + t =map_sysmem (*current, len);memset (t,0 ,sizeof (struct smbios_type0 ));fill_smbios_header (t,SMBIOS_BIOS_INFORMATION , len, handle); t->vendor =smbios_add_string (t->eos,"U-Boot" );
@@ -101,16 +103,18 @@ static int smbios_write_type0(ulong *current, int handle) len = t->length +smbios_string_table_len (t->eos); *current += len; +unmap_sysmem (t);return len; }static int smbios_write_type1 (ulong *current,int handle) { -struct smbios_type1 *t = (struct smbios_type1 *)*current; +struct smbios_type1 *t;int len =sizeof (struct smbios_type1 );char *serial_str =env_get ("serial#" ); + t =map_sysmem (*current, len);memset (t,0 ,sizeof (struct smbios_type1 ));fill_smbios_header (t,SMBIOS_SYSTEM_INFORMATION , len, handle); t->manufacturer =smbios_add_string (t->eos,CONFIG_SMBIOS_MANUFACTURER );
@@ -122,15 +126,17 @@ static int smbios_write_type1(ulong *current, int handle) len = t->length +smbios_string_table_len (t->eos); *current += len; +unmap_sysmem (t);return len; }static int smbios_write_type2 (ulong *current,int handle) { -struct smbios_type2 *t = (struct smbios_type2 *)*current; +struct smbios_type2 *t;int len =sizeof (struct smbios_type2 ); + t =map_sysmem (*current, len);memset (t,0 ,sizeof (struct smbios_type2 ));fill_smbios_header (t,SMBIOS_BOARD_INFORMATION , len, handle); t->manufacturer =smbios_add_string (t->eos,CONFIG_SMBIOS_MANUFACTURER );
@@ -140,15 +146,17 @@ static int smbios_write_type2(ulong *current, int handle) len = t->length +smbios_string_table_len (t->eos); *current += len; +unmap_sysmem (t);return len; }static int smbios_write_type3 (ulong *current,int handle) { -struct smbios_type3 *t = (struct smbios_type3 *)*current; +struct smbios_type3 *t;int len =sizeof (struct smbios_type3 ); + t =map_sysmem (*current, len);memset (t,0 ,sizeof (struct smbios_type3 ));fill_smbios_header (t,SMBIOS_SYSTEM_ENCLOSURE , len, handle); t->manufacturer =smbios_add_string (t->eos,CONFIG_SMBIOS_MANUFACTURER );
@@ -160,6 +168,7 @@ static int smbios_write_type3(ulong *current, int handle) len = t->length +smbios_string_table_len (t->eos); *current += len; +unmap_sysmem (t);return len; }
@@ -198,9 +207,10 @@ static void smbios_write_type4_dm(struct smbios_type4 *t)static int smbios_write_type4 (ulong *current,int handle) { -struct smbios_type4 *t = (struct smbios_type4 *)*current; +struct smbios_type4 *t;int len =sizeof (struct smbios_type4 ); + t =map_sysmem (*current, len);memset (t,0 ,sizeof (struct smbios_type4 ));fill_smbios_header (t,SMBIOS_PROCESSOR_INFORMATION , len, handle); t->processor_type =SMBIOS_PROCESSOR_TYPE_CENTRAL ;
@@ -214,32 +224,37 @@ static int smbios_write_type4(ulong *current, int handle) len = t->length +smbios_string_table_len (t->eos); *current += len; +unmap_sysmem (t);return len; }static int smbios_write_type32 (ulong *current,int handle) { -struct smbios_type32 *t = (struct smbios_type32 *)*current; +struct smbios_type32 *t;int len =sizeof (struct smbios_type32 ); + t =map_sysmem (*current, len);memset (t,0 ,sizeof (struct smbios_type32 ));fill_smbios_header (t,SMBIOS_SYSTEM_BOOT_INFORMATION , len, handle); *current += len; +unmap_sysmem (t);return len; }static int smbios_write_type127 (ulong *current,int handle) { -struct smbios_type127 *t = (struct smbios_type127 *)*current; +struct smbios_type127 *t;int len =sizeof (struct smbios_type127 ); + t =map_sysmem (*current, len);memset (t,0 ,sizeof (struct smbios_type127 ));fill_smbios_header (t,SMBIOS_END_OF_TABLE , len, handle); *current += len; +unmap_sysmem (t);return len; }
@@ -268,7 +283,7 @@ ulong write_smbios_table(ulong addr)/* 16 byte align the table address */ addr =ALIGN (addr,16 ); - se = (struct smbios_entry *)(uintptr_t )addr; + se =map_sysmem (addr,sizeof (struct smbios_entry ));memset (se,0 ,sizeof (struct smbios_entry )); addr +=sizeof (struct smbios_entry );
@@ -297,6 +312,7 @@ ulong write_smbios_table(ulong addr) isize =sizeof (struct smbios_entry ) -SMBIOS_INTERMEDIATE_OFFSET ; se->intermediate_checksum =table_compute_checksum (istart, isize); se->checksum =table_compute_checksum (se,sizeof (struct smbios_entry )); +unmap_sysmem (se);return addr; }
Signature · 1 line
2.18.0.rc1.244.gcf134e6275-goog
PATCH v6 5/13 efi: sandbox: Add distroboot support
In reply to Simon GlassWith sandbox these values depend on the host system. Let's assume that it is x86_64 for now.
---
Changes in v6:
- Warn about building sandbox EFI support on something other than __x86_64__
Changes in v5: None
Changes in v4: None
Changes in v3: None
Changes in v2: None
include/config_distro_bootcmd.h | 13 +++++++++++++ 1 file changed, 13 insertions(+)
include/config_distro_bootcmd.h +13 -0
@@ -251,6 +251,8 @@#elif defined(CONFIG_ARM )#define BOOTENV_EFI_PXE_ARCH "0xa"#define BOOTENV_EFI_PXE_VCI "PXEClient:Arch:00010:UNDI:003000" + +/* For sandbox we only support 64-bit x86 at present */ #elif defined(CONFIG_X86 )/* Always assume we're running 64bit */ #define BOOTENV_EFI_PXE_ARCH "0x7"
@@ -261,6 +263,17 @@#elif defined(CONFIG_CPU_RISCV_64 )#define BOOTENV_EFI_PXE_ARCH "0x1b"#define BOOTENV_EFI_PXE_VCI "PXEClient:Arch:00027:UNDI:003000" +#elif defined(CONFIG_SANDBOX ) +/* +* TODO(sjg at chromium.org): Consider providing a way to enable sandbox features +* based on the host architecture +*/ +# ifndef __x86_64__ +# warning "sandbox EFI support is only tested on 64-bit x86" +# endif +/* To support other *host* architectures this should be changed */ +#define BOOTENV_EFI_PXE_ARCH "0x7" +#define BOOTENV_EFI_PXE_VCI "PXEClient:Arch:00007:UNDI:003000"#else #error Please specify an EFI client identifier#endif
Signature · 1 line
2.18.0.rc1.244.gcf134e6275-goog
PATCH v6 5/13 efi: sandbox: Add distroboot support
Quoted 22 lines · Simon Glass
With sandbox these values depend on the host system. Let's assume that it is
x86_64for now.Signed-off-by Simon Glass <[email protected]>---
Changes in v6:
- Warn about building sandbox EFI support on something other than__x86_64__Changes in v5: None
Changes in v4: None
Changes in v3: None
Changes in v2: Noneinclude/config_distro_bootcmd.h | 13 +++++++++++++ 1 file changed, 13 insertions(+)include/config_distro_bootcmd.h +0
-0
include/config_distro_bootcmd.h line 255
Quoted 5 lines
include/config_distro_bootcmd.h +2
-0@@ -251,6 +251,8 @@#elif defined(CONFIG_ARM )#define BOOTENV_EFI_PXE_ARCH "0xa"#define BOOTENV_EFI_PXE_VCI "PXEClient:Arch:00010:UNDI:003000" + +/* For sandbox we only support 64-bit x86 at present */
I think the TODO in the CONFIG_SANDBOX part is enough, no need to duplicate it here.
Alex
Quoted 21 lines
include/config_distro_bootcmd.h +11
-0#elif defined(CONFIG_X86 )/* Always assume we're running 64bit */ #define BOOTENV_EFI_PXE_ARCH "0x7"@@ -261,6 +263,17 @@#elif defined(CONFIG_CPU_RISCV_64 )#define BOOTENV_EFI_PXE_ARCH "0x1b"#define BOOTENV_EFI_PXE_VCI "PXEClient:Arch:00027:UNDI:003000" +#elif defined(CONFIG_SANDBOX ) +/* +* TODO(sjg at chromium.org): Consider providing a way to enable sandbox features +* based on the host architecture +*/ +# ifndef __x86_64__ +# warning "sandbox EFI support is only tested on 64-bit x86" +# endif +/* To support other *host* architectures this should be changed */ +#define BOOTENV_EFI_PXE_ARCH "0x7" +#define BOOTENV_EFI_PXE_VCI "PXEClient:Arch:00007:UNDI:003000"#else #error Please specify an EFI client identifier#endif
PATCH v6 6/13 efi: sandbox: Add relocation constants
In reply to Simon GlassAdd these so that we can build the EFI loader for sandbox. The values are for x86_64 so potentially bogus. But we don't support relocation within sandbox anyway.
---
Changes in v6:
- Warn about building sandbox EFI support on something other than __x86_64__
Changes in v5: None
Changes in v4: None
Changes in v3: None
Changes in v2: None
lib/efi_loader/efi_runtime.c | 12 ++++++++++++ 1 file changed, 12 insertions(+)
lib/efi_loader/efi_runtime.c +12 -0
@@ -58,6 +58,18 @@ struct dyn_sym {#define R_ABSOLUTE R_RISCV_64#define SYM_INDEX 32#endif + +/* For sandbox we only support 64-bit x86 at present */ +#elif defined(CONFIG_SANDBOX ) +/* +* TODO(sjg at chromium.org): Consider providing a way to enable sandbox features +* based on the host architecture +*/ +# ifndef __x86_64__ +# warning "sandbox EFI support is only tested on 64-bit x86" +# endif +#define R_RELATIVE 8 +#define R_MASK 0xffffffffULL#else #error Need to add relocation awareness#endif
Signature · 1 line
2.18.0.rc1.244.gcf134e6275-goog
PATCH v6 6/13 efi: sandbox: Add relocation constants
Quoted 23 lines · Simon Glass
Add these so that we can build the EFI loader for sandbox. The values are for
x86_64so potentially bogus. But we don't support relocation within sandbox anyway.Signed-off-by Simon Glass <[email protected]>---
Changes in v6:
- Warn about building sandbox EFI support on something other than__x86_64__Changes in v5: None
Changes in v4: None
Changes in v3: None
Changes in v2: Nonelib/efi_loader/efi_runtime.c | 12 ++++++++++++ 1 file changed, 12 insertions(+)lib/efi_loader/efi_runtime.c +0
-0
lib/efi_loader/efi_runtime.c struct dyn_sym { · line 62
Quoted 5 lines
lib/efi_loader/efi_runtime.c +2
-0@@ -58,6 +58,18 @@ struct dyn_sym {#define R_ABSOLUTE R_RISCV_64#define SYM_INDEX 32#endif + +/* For sandbox we only support 64-bit x86 at present */
Same here. TODO with #ifndef is plenty to tell people what you're trying to say :)
Alex
Quoted 13 lines
lib/efi_loader/efi_runtime.c +10
-0+#elif defined(CONFIG_SANDBOX ) +/* +* TODO(sjg at chromium.org): Consider providing a way to enable sandbox features +* based on the host architecture +*/ +# ifndef __x86_64__ +# warning "sandbox EFI support is only tested on 64-bit x86" +# endif +#define R_RELATIVE 8 +#define R_MASK 0xffffffffULL#else #error Need to add relocation awareness#endif
PATCH v6 7/13 efi: Add a comment about duplicated ELF constants
In reply to Simon GlassThese constants are defined in arch-specific code but redefined here. Add a TODO to clean this up.
---
Changes in v6: None
Changes in v5:
- Drop period after "elf" in comment
Changes in v4: None
Changes in v3: None
Changes in v2: None
lib/efi_loader/efi_runtime.c | 4 ++++ 1 file changed, 4 insertions(+)
lib/efi_loader/efi_runtime.c +4 -0
@@ -28,6 +28,10 @@ static efi_status_t __efi_runtime EFIAPI efi_unimplemented(void); static efi_status_t __efi_runtime EFIAPI efi_device_error(void); static efi_status_t __efi_runtime EFIAPI efi_invalid_parameter(void); +/* +* TODO(sjg at chromium.org): These defines and structs should come from the elf +* header for each arch (or a generic header) rather than being repeated here. +*/ #if defined(CONFIG_ARM64 )#define R_RELATIVE 1027#define R_MASK 0xffffffffULL
Signature · 1 line
2.18.0.rc1.244.gcf134e6275-goog
PATCH v6 7/13 efi: Add a comment about duplicated ELF constants
Quoted 5 lines · Simon Glass
These constants are defined in arch-specific code but redefined here. Add a TODO to clean this up.
Signed-off-by Simon Glass <[email protected]>Reviewed-by Heinrich Schuchardt <[email protected]>
This patch is part of efi-next already.
Alex
PATCH v6 8/13 efi: sandbox: Enable EFI loader builder for sandbox
In reply to Simon GlassThis allows this feature to build within sandbox. This is for testing purposes only since it is not possible for sandbox to load native code.
---
Changes in v6: None
Changes in v5: None
Changes in v4: None
Changes in v3:
- Init the 'rows' and 'cols' vars to avoid a compiler error (gcc 4.8.4)
Changes in v2: None
lib/efi_loader/Kconfig | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
lib/efi_loader/Kconfig +1 -1
@@ -1,6 +1,6 @@ config EFI_LOADER bool "Support running EFI Applications in U-Boot" - depends on (ARM || X86 || RISCV) && OF_LIBFDT + depends on (ARM || X86 || RISCV || SANDBOX) && OF_LIBFDT # We do not support bootefi booting ARMv7 in non-secure mode depends on !ARMV7_NONSEC # We need EFI_STUB_64BIT to be set on x86_64 with EFI_STUB
Signature · 1 line
2.18.0.rc1.244.gcf134e6275-goog
PATCH v6 8/13 efi: sandbox: Enable EFI loader builder for sandbox
Quoted 4 lines · Simon Glass
This allows this feature to build within sandbox. This is for testing purposes only since it is not possible for sandbox to load native code.
Signed-off-by Simon Glass <[email protected]>
Are you sure you want to enable a "builder", not the "build"? (see subject line)
Alex
PATCH v6 9/13 efi: Split out test init/uninit into functions
In reply to Simon GlassWe plan to run more than one EFI test. In order to avoid duplicating code, create functions which handle preparing for running the test and cleaning up afterwards.
Also shorten the awfully long variable names here.
---
Changes in v6: None
Changes in v5:
- Drop call to efi_init_obj_list() which is now done in do_bootefi()
Changes in v4: None
Changes in v3:
- Add new patch to split out test init/uninit into functions
Changes in v2: None
cmd/bootefi.c | 87 +++++++++++++++++++++++++++++++++++++-------------- 1 file changed, 63 insertions(+), 24 deletions(-)
cmd/bootefi.c +63 -24
@@ -347,6 +347,60 @@ exit:return ret; } +#ifdef CONFIG_CMD_BOOTEFI_SELFTEST +/** +* bootefi_test_prepare() - prepare to run an EFI test +* +* This sets things up so we can call EFI functions. This involves preparing +* the 'gd' pointer and setting up the load ed image data structures. +* +* @image: Pointer to a struct which will hold the loaded image info +* @obj: Pointer to a struct which will hold the loaded image object +* @path: File path to the test being run (often just the test name with a +* backslash before it +* @test_func: Address of the test function that is being run +* @return 0 if OK, -ve on error +*/ +static efi_status_t bootefi_test_prepare (struct efi_loaded_image *image, +struct efi_object *obj, +const char *path,ulong test_func) +{ +memset (image,'\0' ,sizeof (*image)); +memset (obj,'\0' ,sizeof (*obj)); +/* Construct a dummy device path */ + bootefi_device_path =efi_dp_from_mem (EFI_RESERVED_MEMORY_TYPE , + (uintptr_t )test_func, + (uintptr_t )test_func); + bootefi_image_path =efi_dp_from_file (NULL ,0 , path); +efi_setup_loaded_image (image, obj, bootefi_device_path, + bootefi_image_path); +/* +* gd lives in a fixed register which may get clobbered while we execute +* the payload. So save it here and restore it on every callback entry +*/ +efi_save_gd (); + +/* Transfer environment variable efi_selftest as load options */ +set_load_options (image,"efi_selftest" ); + +return 0 ; +} + +/** +* bootefi_test_finish() - finish up after running an EFI test +* +* @image: Pointer to a struct which holds the loaded image info +* @obj: Pointer to a struct which holds the loaded image object +*/ +static void bootefi_test_finish (struct efi_loaded_image *image, +struct efi_object *obj) +{ +efi_restore_gd (); +free (image->load_options); +list_del (&obj->link); +} +#endif /* CONFIG_CMD_BOOTEFI_SELFTEST */ +static int do_bootefi_bootmgr_exec (void ) {struct efi_device_path *device_path, *file_path;
@@ -425,31 +479,16 @@ static int do_bootefi(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[])#endif #ifdef CONFIG_CMD_BOOTEFI_SELFTEST if (!strcmp (argv[1 ],"selftest" )) { -struct efi_loaded_image loaded_image_info = {}; -struct efi_object loaded_image_info_obj = {}; - -/* Construct a dummy device path. */ - bootefi_device_path =efi_dp_from_mem (EFI_RESERVED_MEMORY_TYPE , - (uintptr_t )&efi_selftest, - (uintptr_t )&efi_selftest); - bootefi_image_path =efi_dp_from_file (NULL ,0 ,"\\selftest" ); - -efi_setup_loaded_image (&loaded_image_info, - &loaded_image_info_obj, - bootefi_device_path, bootefi_image_path); -/* -* gd lives in a fixed register which may get clobbered while we -* execute the payload. So save it here and restore it on every -* callback entry -*/ -efi_save_gd (); -/* Transfer environment variable efi_selftest as load options */ -set_load_options (&loaded_image_info,"efi_selftest" ); +struct efi_loaded_image image; +struct efi_object obj; + +if (bootefi_test_prepare (&image, &obj,"\\selftest" , + (uintptr_t )&efi_selftest)) +return CMD_RET_FAILURE ; +/* Execute the test */ - r =efi_selftest (loaded_image_info_obj.handle, &systab); -efi_restore_gd (); -free (loaded_image_info.load_options); -list_del (&loaded_image_info_obj.link); + r =efi_selftest (obj.handle, &systab); +bootefi_test_finish (&image, &obj);return r !=EFI_SUCCESS ; } else#endif
Signature · 1 line
2.18.0.rc1.244.gcf134e6275-goog
PATCH v6 10/13 efi: sandbox: Add a simple 'bootefi test' command
In reply to Simon GlassThis jumps to test code which can call directly into the EFI support. It does not need a separate image so it is easy to write tests with it.
This test can be executed without causing problems to the run-time environemnt (e.g. U-Boot does not need to reboot afterwards).
For now the test just outputs a message. To try it:
./sandbox/u-boot -c "bootefi test"
U-Boot 2017.09-00204-g696c9855fe (Sep 17 2017 - 16:43:53 -0600)
DRAM: 128 MiB
MMC:
Using default environment
In: serial Out: serial Err: serial SCSI: Net: No ethernet found. IDE: Bus 0: not available
Found 0 disks
WARNING: booting without device tree
Hello, world!
Test passed
---
Changes in v6: None
Changes in v5: None
Changes in v4:
- Rebase to master
- Update SPDX tags
Changes in v3:
- Rebase to master
Changes in v2:
- Rebase to master
cmd/bootefi.c | 26 ++++++++++++++++++++------ include/efi_loader.h | 3 +++ lib/efi_loader/Kconfig | 10 ++++++++++ lib/efi_loader/Makefile | 1 + lib/efi_loader/efi_test.c | 16 ++++++++++++++++ 5 files changed, 50 insertions(+), 6 deletions(-) create mode 100644 lib/efi_loader/efi_test.c
cmd/bootefi.c +20 -6
@@ -347,7 +347,6 @@ exit:return ret; } -#ifdef CONFIG_CMD_BOOTEFI_SELFTEST /** * bootefi_test_prepare() - prepare to run an EFI test *
@@ -399,7 +398,6 @@ static void bootefi_test_finish(struct efi_loaded_image *image,free (image->load_options);list_del (&obj->link); } -#endif /* CONFIG_CMD_BOOTEFI_SELFTEST */ static int do_bootefi_bootmgr_exec (void ) {
@@ -431,8 +429,10 @@ static int do_bootefi_bootmgr_exec(void)/* Interpreter command to boot an arbitrary EFI image from memory */ static int do_bootefi (cmd_tbl_t *cmdtp,int flag,int argc,char *const argv[]) { -unsigned long addr; +struct efi_loaded_image image; +struct efi_object obj;char *saddr; +unsigned long addr;efi_status_t r;void *fdt_addr;
@@ -477,11 +477,25 @@ static int do_bootefi(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[])memcpy ((char *)addr, __efi_helloworld_begin, size); } else#endif +if (IS_ENABLED (CONFIG_BOOTEFI_TEST ) && !strcmp (argv[1 ],"test" )) { +int ret; + +if (bootefi_test_prepare (&image, &obj,"\\test" , + (ulong)&efi_test)) +return CMD_RET_FAILURE ; + + ret =efi_test (&image, &systab); +bootefi_test_finish (&image, &obj); +if (ret) { +printf ("Test failed: err=%d\n" , ret); +return CMD_RET_FAILURE ; + } +printf ("Test passed\n" ); + +return 0 ; + }#ifdef CONFIG_CMD_BOOTEFI_SELFTEST if (!strcmp (argv[1 ],"selftest" )) { -struct efi_loaded_image image; -struct efi_object obj; -if (bootefi_test_prepare (&image, &obj,"\\selftest" , (uintptr_t )&efi_selftest))return CMD_RET_FAILURE ;
include/efi_loader.h +3 -0
@@ -442,6 +442,9 @@ efi_status_t EFIAPI efi_set_variable(u16 *variable_name, efi_guid_t *vendor,void *efi_bootmgr_load (struct efi_device_path **device_path,struct efi_device_path **file_path); +/* Perform EFI tests */ +int efi_test (efi_handle_t image_handle,struct efi_system_table *systab); +#else /* defined(EFI_LOADER) && !defined(CONFIG_SPL_BUILD) */ /* Without CONFIG_EFI_LOADER we don't have a runtime section, stub it out */
lib/efi_loader/Kconfig +10 -0
@@ -25,3 +25,13 @@ config EFI_LOADER_BOUNCE_BUFFER Some hardware does not support DMA to full 64bit addresses. For this hardware we can create a bounce buffer so that payloads don't have to worry about platform details. + +config BOOTEFI_TEST + bool "Provide a test for the EFI loader" + depends on EFI_LOADER && SANDBOX + default y + help + Provides a test of the EFI loader functionality accessed via the + command line ('bootefi test'). This runs within U-Boot so does not + need a separate EFI application to work. It aims to include coverage + of all EFI code which can be accessed within sandbox.
lib/efi_loader/Makefile +1 -0
@@ -23,3 +23,4 @@ obj-$(CONFIG_DM_VIDEO) += efi_gop.o obj-$(CONFIG_PARTITIONS) += efi_disk.o obj-$(CONFIG_NET) += efi_net.o obj-$(CONFIG_GENERATE_SMBIOS_TABLE) += efi_smbios.o +obj-$(CONFIG_BOOTEFI_TEST) += efi_test.o
lib/efi_loader/efi_test.c new file +16 -0
@@ -0,0 +1,16 @@ +// SPDX-License-Identifier: GPL-2.0+ +/* +* Copyright (c) 2017, Google Inc. +*/ + +#include <common.h> +#include <efi_api.h> + +int efi_test (efi_handle_t image_handle,struct efi_system_table *systable) +{ +struct efi_simple_text_output_protocol *con_out = systable->con_out; + + con_out->output_string (con_out,L"Hello, world!\n" ); + +return 0 ; +}
Signature · 1 line
2.18.0.rc1.244.gcf134e6275-goog
PATCH v6 11/13 efi: Create a function to set up for running EFI code
In reply to Simon GlassAdd a new bootefi_run_prepare() function which holds common code used to set up U-Boot to run EFI code. Make use of this from the existing bootefi_test_prepare() function, as well as do_bootefi_exec().
Also shorten a few variable names.
---
Changes in v6: None
Changes in v5:
- Drop call to efi_init_obj_list() which is now done in do_bootefi()
- Introduce load_options_path to specifyc U-Boot env var for load_options_path
Changes in v4:
- Rebase to master
Changes in v3:
- Add patch to create a function to set up for running EFI code
Changes in v2: None
cmd/bootefi.c | 79 ++++++++++++++++++++++++++++----------------------- 1 file changed, 43 insertions(+), 36 deletions(-)
cmd/bootefi.c +43 -36
@@ -253,6 +253,26 @@ static efi_status_t efi_install_fdt(void *fdt)return ret; } +static efi_status_t bootefi_run_prepare (struct efi_loaded_image *image, +struct efi_object *obj, +const char *load_options_path, +struct efi_device_path *device_path, +struct efi_device_path *image_path) +{ +efi_setup_loaded_image (image, obj, device_path, image_path); + +/* +* gd lives in a fixed register which may get clobbered while we execute +* the payload. So save it here and restore it on every callback entry +*/ +efi_save_gd (); + +/* Transfer environment variable as load options */ +set_load_options (image, load_options_path); + +return 0 ; +} +/* * Load an EFI payload into a newly allocated piece of memory, register all * EFI objects it would want to access and jump to it.
@@ -261,8 +281,8 @@ static efi_status_t do_bootefi_exec(void *efi,struct efi_device_path *device_path, struct efi_device_path *image_path) { - struct efi_loaded_image loaded_image_info = {}; -struct efi_object loaded_image_info_obj = {}; + struct efi_loaded_image image; +struct efi_object obj;struct efi_device_path *memdp =NULL ;efi_status_t ret;
@@ -283,19 +303,13 @@ static efi_status_t do_bootefi_exec(void *efi,assert (device_path && image_path); } -efi_setup_loaded_image (&loaded_image_info, &loaded_image_info_obj, - device_path, image_path); + ret =bootefi_run_prepare (&image, &obj,"bootargs" , device_path, + image_path); +if (ret) +return ret; -/* -* gd lives in a fixed register which may get clobbered while we execute -* the payload. So save it here and restore it on every callback entry -*/ -efi_save_gd (); - -/* Transfer environment variable bootargs as load options */ -set_load_options (&loaded_image_info,"bootargs" );/* Load the EFI payload */ - entry =efi_load_pe (efi, &loaded_image_info); + entry =efi_load_pe (efi, &image);if (!entry) { ret =EFI_LOAD_ERROR ; goto exit;
@@ -303,10 +317,10 @@ static efi_status_t do_bootefi_exec(void *efi,if (memdp) {struct efi_device_path_memory *mdp = (void *)memdp; - mdp->memory_type = loaded_image_info.image_code_type; - mdp->start_address = (uintptr_t )loaded_image_info.image_base; + mdp->memory_type = image.image_code_type; + mdp->start_address = (uintptr_t )image.image_base; mdp->end_address = mdp->start_address + - loaded_image_info.image_size; + image.image_size; }/* we don't support much: */
@@ -316,8 +330,8 @@ static efi_status_t do_bootefi_exec(void *efi,/* Call our payload! */ debug ("%s:%d Jumping to 0x%lx\n" , __func__, __LINE__, (long )entry); -if (setjmp (&loaded_image_info.exit_jmp)) { - ret = loaded_image_info.exit_status; +if (setjmp (&image.exit_jmp)) { + ret = image.exit_status; goto exit; }
@@ -329,7 +343,7 @@ static efi_status_t do_bootefi_exec(void *efi,/* Move into EL2 and keep running there */ armv8_switch_to_el2 ((ulong )entry, - (ulong)&loaded_image_info_obj.handle, + (ulong)&obj.handle, (ulong)&systab,0 , (ulong )efi_run_in_el2,ES_TO_AARCH64 );
@@ -338,11 +352,11 @@ static efi_status_t do_bootefi_exec(void *efi, }#endif - ret =efi_do_enter (loaded_image_info_obj.handle, &systab, entry); + ret =efi_do_enter (obj.handle, &systab, entry); exit:/* image has returned, loaded-image obj goes *poof*: */ -list_del (&loaded_image_info_obj.link); +list_del (&obj.link);return ret; }
@@ -358,11 +372,13 @@ exit:* @path: File path to the test being run (often just the test name with a * backslash before it * @test_func: Address of the test function that is being run +* @load_options_path: U-Boot environment variable to use as load options * @return 0 if OK, -ve on error */ static efi_status_t bootefi_test_prepare (struct efi_loaded_image *image,struct efi_object *obj, -const char *path,ulong test_func) +const char *path,ulong test_func, +const char *load_options_path) {memset (image,'\0' ,sizeof (*image));memset (obj,'\0' ,sizeof (*obj));
@@ -371,18 +387,8 @@ static efi_status_t bootefi_test_prepare(struct efi_loaded_image *image, (uintptr_t )test_func, (uintptr_t)test_func); bootefi_image_path =efi_dp_from_file (NULL ,0 , path); -efi_setup_loaded_image (image, obj, bootefi_device_path, - bootefi_image_path); -/* -* gd lives in a fixed register which may get clobbered while we execute -* the payload. So save it here and restore it on every callback entry -*/ -efi_save_gd (); - -/* Transfer environment variable efi_selftest as load options */ -set_load_options (image,"efi_selftest" ); - -return 0 ; +return bootefi_run_prepare (image, obj, load_options_path, + bootefi_device_path, bootefi_image_path); }/**
@@ -481,7 +487,7 @@ static int do_bootefi(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[])int ret;if (bootefi_test_prepare (&image, &obj,"\\test" , - (ulong)&efi_test)) + (ulong)&efi_test,"efi_test" ))return CMD_RET_FAILURE ; ret =efi_test (&image, &systab);
@@ -497,7 +503,8 @@ static int do_bootefi(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[])#ifdef CONFIG_CMD_BOOTEFI_SELFTEST if (!strcmp (argv[1 ],"selftest" )) {if (bootefi_test_prepare (&image, &obj,"\\selftest" , - (uintptr_t )&efi_selftest)) + (uintptr_t )&efi_selftest, +"efi_selftest" ))return CMD_RET_FAILURE ;/* Execute the test */
Signature · 1 line
2.18.0.rc1.244.gcf134e6275-goog
PATCH v6 12/13 efi: Rename bootefi_test_finish() to bootefi_run_finish()
In reply to Simon GlassThis function can be used from do_bootefi_exec() so that we use mostly the same code for a normal EFI application and an EFI test.
Rename the function and use it in both places.
---
Changes in v6: None
Changes in v5:
- Rebase to master
Changes in v4:
- Rebase to master
Changes in v3:
- Add new patch to rename bootefi_test_finish() to bootefi_run_finish()
Changes in v2: None
cmd/bootefi.c | 35 +++++++++++++++++------------------ 1 file changed, 17 insertions(+), 18 deletions(-)
cmd/bootefi.c +17 -18
@@ -273,6 +273,20 @@ static efi_status_t bootefi_run_prepare(struct efi_loaded_image *image,return 0 ; } +/** +* bootefi_run_finish() - finish up after running an EFI test +* +* @image: Pointer to a struct which holds the loaded image info +* @obj: Pointer to a struct which holds the loaded image object +*/ +static void bootefi_run_finish (struct efi_loaded_image *image, +struct efi_object *obj) +{ +efi_restore_gd (); +free (image->load_options); +list_del (&obj->link); +} +/* * Load an EFI payload into a newly allocated piece of memory, register all * EFI objects it would want to access and jump to it.
@@ -355,8 +369,7 @@ static efi_status_t do_bootefi_exec(void *efi, ret =efi_do_enter (obj.handle, &systab, entry); exit: -/* image has returned, loaded-image obj goes *poof*: */ -list_del (&obj.link); +bootefi_run_finish (&image, &obj);return ret; }
@@ -391,20 +404,6 @@ static efi_status_t bootefi_test_prepare(struct efi_loaded_image *image, bootefi_device_path, bootefi_image_path); } -/** -* bootefi_test_finish() - finish up after running an EFI test -* -* @image: Pointer to a struct which holds the loaded image info -* @obj: Pointer to a struct which holds the loaded image object -*/ -static void bootefi_test_finish (struct efi_loaded_image *image, -struct efi_object *obj) -{ -efi_restore_gd (); -free (image->load_options); -list_del (&obj->link); -} -static int do_bootefi_bootmgr_exec (void ) {struct efi_device_path *device_path, *file_path;
@@ -491,7 +490,7 @@ static int do_bootefi(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[])return CMD_RET_FAILURE ; ret =efi_test (&image, &systab); -bootefi_test_finish (&image, &obj); +bootefi_run_finish (&image, &obj);if (ret) {printf ("Test failed: err=%d\n" , ret);return CMD_RET_FAILURE ;
@@ -509,7 +508,7 @@ static int do_bootefi(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[])/* Execute the test */ r =efi_selftest (obj.handle, &systab); -bootefi_test_finish (&image, &obj); +bootefi_run_finish (&image, &obj);return r !=EFI_SUCCESS ; } else#endif
Signature · 1 line
2.18.0.rc1.244.gcf134e6275-goog
PATCH v6 13/13 Revert "buildman: Extract environment as part of each build"
In reply to Simon GlassThis reverts commit 0ddc510ea37aa578b8cb197840a5125409978bec.
---
Changes in v6: None
Changes in v5: None
Changes in v4: None
Changes in v3: None
Changes in v2: None
tools/buildman/builderthread.py | 10 ---------- tools/buildman/func_test.py | 5 ----- 2 files changed, 15 deletions(-)
tools/buildman/builderthread.py +0 -10
@@ -351,16 +351,6 @@ class BuilderThread(threading.Thread): lines.append(size_result.stdout.splitlines()[1] + ' ' + rodata_size) - # Extract the environment from U-Boot and dump it out - cmd = ['%sobjcopy' % self.toolchain.cross, '-O', 'binary', - '-j', '.rodata.default_environment', - 'env/built-in.o', 'uboot.env'] - command.RunPipe([cmd], capture=True, - capture_stderr=True, cwd=result.out_dir, - raise_on_error=False, env=env) - ubootenv = os.path.join(result.out_dir, 'uboot.env') - self.CopyFiles(result.out_dir, build_dir, '', ['uboot.env']) - # Write out the image sizes file. This is similar to the output # of binutil's 'size' utility, but it omits the header line and # adds an additional hex value at the end of each line for the
tools/buildman/func_test.py +0 -5
@@ -327,9 +327,6 @@ class TestFunctional(unittest.TestCase): def _HandleCommandObjdump(self, args): return command.CommandResult(return_code=0) - def _HandleCommandObjcopy(self, args): - return command.CommandResult(return_code=0) - def _HandleCommandSize(self, args): return command.CommandResult(return_code=0)
@@ -362,8 +359,6 @@ class TestFunctional(unittest.TestCase): return self._HandleCommandNm(args) elif cmd.endswith('objdump'): return self._HandleCommandObjdump(args) - elif cmd.endswith('objcopy'): - return self._HandleCommandObjcopy(args) elif cmd.endswith( 'size'): return self._HandleCommandSize(args)
Signature · 1 line
2.18.0.rc1.244.gcf134e6275-goog
PATCH v6 13/13 Revert "buildman: Extract environment as part of each build"
Hi,
Quoted 5 lines · Simon Glass
This reverts commit
0ddc510ea37aa578b8cb197840a5125409978bec.Signed-off-by Simon Glass <[email protected]>---
Sorry, please ignore this patch. I am trying to track down a buildman issue and put this commit on top just before sending :-(
Regards,
Simon