coreboot-kgpe-d16/payloads/libpayload/arch/arm64/mmu.c

746 lines
20 KiB
C
Raw Normal View History

/*
* This file is part of the coreboot project.
*
* Copyright 2014 Google Inc.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. The name of the author may not be used to endorse or promote products
* derived from this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*/
#include <assert.h>
#include <stdlib.h>
#include <stdint.h>
#include <string.h>
#include <arch/mmu.h>
#include <arch/lib_helpers.h>
#include <arch/cache.h>
/* Maximum number of XLAT Tables available based on ttb buffer size */
static unsigned int max_tables;
/* Address of ttb buffer */
static uint64_t *xlat_addr;
static int free_idx;
static uint8_t ttb_buffer[TTB_DEFAULT_SIZE] __attribute__((aligned(GRANULE_SIZE)));
arm64: Decouple MMU functions from memranges The current arm64 MMU interface is difficult to use in pre-RAM environments. It is based on the memranges API which makes use of malloc(), and early stages usually don't have a heap. It is also built as a one-shot interface that requires all memory ranges to be laid out beforehand, which is a problem when existing areas need to change (e.g. after initializing DRAM). The long-term goal of this patch is to completely switch to a configure-as-you-go interface based on the mmu_config_range() function, similar to what ARM32 does. As a first step this feature is added side-by-side to the existing interface so that existing SoC implementations continue to work and can be slowly ported over one by one. Like the ARM32 version it does not garbage collect page tables that become unused, so repeated mapping at different granularities will exhaust the available table space (this is presumed to be a reasonable limitation for a firmware environment and keeps the code much simpler). Also do some cleanup, align comments between coreboot and libpayload for easier diffing, and change all error cases to assert()s. Right now the code just propagates error codes up the stack until it eventually reaches a function that doesn't check them anymore. MMU configuration errors (essentially just misaligned requests and running out of table space) should always be compile-time programming errors, so failing hard and fast seems like the best way to deal with them. BRANCH=None BUG=None TEST=Compile-tested rush_ryu. Booted on Oak and hacked MMU init to use mmu_config_range() insted of memranges. Confirmed that CRCs over all page tables before and after the change are equal. Change-Id: I93585b44a277c1d96d31ee9c3dd2522b5e10085b Signed-off-by: Patrick Georgi <pgeorgi@chromium.org> Original-Commit-Id: f10fcba107aba1f3ea239471cb5a4f9239809539 Original-Change-Id: I6a2a11e3b94e6ae9e1553871f0cccd3b556b3e65 Original-Signed-off-by: Julius Werner <jwerner@chromium.org> Original-Reviewed-on: https://chromium-review.googlesource.com/271991 Original-Reviewed-by: Aaron Durbin <adurbin@chromium.org> Reviewed-on: http://review.coreboot.org/10304 Tested-by: build bot (Jenkins) Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net> Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
2015-05-18 22:11:12 +02:00
static const char * const tag_to_string[] = {
[TYPE_NORMAL_MEM] = "normal",
[TYPE_DEV_MEM] = "device",
[TYPE_DMA_MEM] = "uncached",
};
/*
* The usedmem_ranges is used to describe all the memory ranges that are
* actually used by payload i.e. _start -> _end in linker script and the
* coreboot tables. This is required for two purposes:
* 1) During the pre_sysinfo_scan_mmu_setup, these are the only ranges
* initialized in the page table as we do not know the entire memory map.
* 2) During the post_sysinfo_scan_mmu_setup, these ranges are used to check if
* the DMA buffer is being placed in a sane location and does not overlap any of
* the used mem ranges.
*/
static struct mmu_ranges usedmem_ranges;
static void __attribute__((noreturn)) mmu_error(void)
{
halt();
}
arm64: Decouple MMU functions from memranges The current arm64 MMU interface is difficult to use in pre-RAM environments. It is based on the memranges API which makes use of malloc(), and early stages usually don't have a heap. It is also built as a one-shot interface that requires all memory ranges to be laid out beforehand, which is a problem when existing areas need to change (e.g. after initializing DRAM). The long-term goal of this patch is to completely switch to a configure-as-you-go interface based on the mmu_config_range() function, similar to what ARM32 does. As a first step this feature is added side-by-side to the existing interface so that existing SoC implementations continue to work and can be slowly ported over one by one. Like the ARM32 version it does not garbage collect page tables that become unused, so repeated mapping at different granularities will exhaust the available table space (this is presumed to be a reasonable limitation for a firmware environment and keeps the code much simpler). Also do some cleanup, align comments between coreboot and libpayload for easier diffing, and change all error cases to assert()s. Right now the code just propagates error codes up the stack until it eventually reaches a function that doesn't check them anymore. MMU configuration errors (essentially just misaligned requests and running out of table space) should always be compile-time programming errors, so failing hard and fast seems like the best way to deal with them. BRANCH=None BUG=None TEST=Compile-tested rush_ryu. Booted on Oak and hacked MMU init to use mmu_config_range() insted of memranges. Confirmed that CRCs over all page tables before and after the change are equal. Change-Id: I93585b44a277c1d96d31ee9c3dd2522b5e10085b Signed-off-by: Patrick Georgi <pgeorgi@chromium.org> Original-Commit-Id: f10fcba107aba1f3ea239471cb5a4f9239809539 Original-Change-Id: I6a2a11e3b94e6ae9e1553871f0cccd3b556b3e65 Original-Signed-off-by: Julius Werner <jwerner@chromium.org> Original-Reviewed-on: https://chromium-review.googlesource.com/271991 Original-Reviewed-by: Aaron Durbin <adurbin@chromium.org> Reviewed-on: http://review.coreboot.org/10304 Tested-by: build bot (Jenkins) Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net> Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
2015-05-18 22:11:12 +02:00
/* Func : get_block_attr
* Desc : Get block descriptor attributes based on the value of tag in memrange
* region
*/
static uint64_t get_block_attr(unsigned long tag)
{
uint64_t attr;
/* We should be in EL2(which is non-secure only) or EL1(non-secure) */
attr = BLOCK_NS;
/* Assuming whole memory is read-write */
attr |= BLOCK_AP_RW;
attr |= BLOCK_ACCESS;
switch (tag) {
case TYPE_NORMAL_MEM:
attr |= BLOCK_SH_INNER_SHAREABLE;
attr |= (BLOCK_INDEX_MEM_NORMAL << BLOCK_INDEX_SHIFT);
break;
case TYPE_DEV_MEM:
attr |= BLOCK_INDEX_MEM_DEV_NGNRNE << BLOCK_INDEX_SHIFT;
attr |= BLOCK_XN;
break;
case TYPE_DMA_MEM:
attr |= BLOCK_INDEX_MEM_NORMAL_NC << BLOCK_INDEX_SHIFT;
break;
}
return attr;
}
arm64: Decouple MMU functions from memranges The current arm64 MMU interface is difficult to use in pre-RAM environments. It is based on the memranges API which makes use of malloc(), and early stages usually don't have a heap. It is also built as a one-shot interface that requires all memory ranges to be laid out beforehand, which is a problem when existing areas need to change (e.g. after initializing DRAM). The long-term goal of this patch is to completely switch to a configure-as-you-go interface based on the mmu_config_range() function, similar to what ARM32 does. As a first step this feature is added side-by-side to the existing interface so that existing SoC implementations continue to work and can be slowly ported over one by one. Like the ARM32 version it does not garbage collect page tables that become unused, so repeated mapping at different granularities will exhaust the available table space (this is presumed to be a reasonable limitation for a firmware environment and keeps the code much simpler). Also do some cleanup, align comments between coreboot and libpayload for easier diffing, and change all error cases to assert()s. Right now the code just propagates error codes up the stack until it eventually reaches a function that doesn't check them anymore. MMU configuration errors (essentially just misaligned requests and running out of table space) should always be compile-time programming errors, so failing hard and fast seems like the best way to deal with them. BRANCH=None BUG=None TEST=Compile-tested rush_ryu. Booted on Oak and hacked MMU init to use mmu_config_range() insted of memranges. Confirmed that CRCs over all page tables before and after the change are equal. Change-Id: I93585b44a277c1d96d31ee9c3dd2522b5e10085b Signed-off-by: Patrick Georgi <pgeorgi@chromium.org> Original-Commit-Id: f10fcba107aba1f3ea239471cb5a4f9239809539 Original-Change-Id: I6a2a11e3b94e6ae9e1553871f0cccd3b556b3e65 Original-Signed-off-by: Julius Werner <jwerner@chromium.org> Original-Reviewed-on: https://chromium-review.googlesource.com/271991 Original-Reviewed-by: Aaron Durbin <adurbin@chromium.org> Reviewed-on: http://review.coreboot.org/10304 Tested-by: build bot (Jenkins) Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net> Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
2015-05-18 22:11:12 +02:00
/* Func : table_desc_valid
* Desc : Check if a table entry contains valid desc
*/
static uint64_t table_desc_valid(uint64_t desc)
{
return((desc & TABLE_DESC) == TABLE_DESC);
}
arm64: Decouple MMU functions from memranges The current arm64 MMU interface is difficult to use in pre-RAM environments. It is based on the memranges API which makes use of malloc(), and early stages usually don't have a heap. It is also built as a one-shot interface that requires all memory ranges to be laid out beforehand, which is a problem when existing areas need to change (e.g. after initializing DRAM). The long-term goal of this patch is to completely switch to a configure-as-you-go interface based on the mmu_config_range() function, similar to what ARM32 does. As a first step this feature is added side-by-side to the existing interface so that existing SoC implementations continue to work and can be slowly ported over one by one. Like the ARM32 version it does not garbage collect page tables that become unused, so repeated mapping at different granularities will exhaust the available table space (this is presumed to be a reasonable limitation for a firmware environment and keeps the code much simpler). Also do some cleanup, align comments between coreboot and libpayload for easier diffing, and change all error cases to assert()s. Right now the code just propagates error codes up the stack until it eventually reaches a function that doesn't check them anymore. MMU configuration errors (essentially just misaligned requests and running out of table space) should always be compile-time programming errors, so failing hard and fast seems like the best way to deal with them. BRANCH=None BUG=None TEST=Compile-tested rush_ryu. Booted on Oak and hacked MMU init to use mmu_config_range() insted of memranges. Confirmed that CRCs over all page tables before and after the change are equal. Change-Id: I93585b44a277c1d96d31ee9c3dd2522b5e10085b Signed-off-by: Patrick Georgi <pgeorgi@chromium.org> Original-Commit-Id: f10fcba107aba1f3ea239471cb5a4f9239809539 Original-Change-Id: I6a2a11e3b94e6ae9e1553871f0cccd3b556b3e65 Original-Signed-off-by: Julius Werner <jwerner@chromium.org> Original-Reviewed-on: https://chromium-review.googlesource.com/271991 Original-Reviewed-by: Aaron Durbin <adurbin@chromium.org> Reviewed-on: http://review.coreboot.org/10304 Tested-by: build bot (Jenkins) Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net> Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
2015-05-18 22:11:12 +02:00
/* Func : setup_new_table
* Desc : Get next free table from TTB and set it up to match old parent entry.
*/
arm64: Decouple MMU functions from memranges The current arm64 MMU interface is difficult to use in pre-RAM environments. It is based on the memranges API which makes use of malloc(), and early stages usually don't have a heap. It is also built as a one-shot interface that requires all memory ranges to be laid out beforehand, which is a problem when existing areas need to change (e.g. after initializing DRAM). The long-term goal of this patch is to completely switch to a configure-as-you-go interface based on the mmu_config_range() function, similar to what ARM32 does. As a first step this feature is added side-by-side to the existing interface so that existing SoC implementations continue to work and can be slowly ported over one by one. Like the ARM32 version it does not garbage collect page tables that become unused, so repeated mapping at different granularities will exhaust the available table space (this is presumed to be a reasonable limitation for a firmware environment and keeps the code much simpler). Also do some cleanup, align comments between coreboot and libpayload for easier diffing, and change all error cases to assert()s. Right now the code just propagates error codes up the stack until it eventually reaches a function that doesn't check them anymore. MMU configuration errors (essentially just misaligned requests and running out of table space) should always be compile-time programming errors, so failing hard and fast seems like the best way to deal with them. BRANCH=None BUG=None TEST=Compile-tested rush_ryu. Booted on Oak and hacked MMU init to use mmu_config_range() insted of memranges. Confirmed that CRCs over all page tables before and after the change are equal. Change-Id: I93585b44a277c1d96d31ee9c3dd2522b5e10085b Signed-off-by: Patrick Georgi <pgeorgi@chromium.org> Original-Commit-Id: f10fcba107aba1f3ea239471cb5a4f9239809539 Original-Change-Id: I6a2a11e3b94e6ae9e1553871f0cccd3b556b3e65 Original-Signed-off-by: Julius Werner <jwerner@chromium.org> Original-Reviewed-on: https://chromium-review.googlesource.com/271991 Original-Reviewed-by: Aaron Durbin <adurbin@chromium.org> Reviewed-on: http://review.coreboot.org/10304 Tested-by: build bot (Jenkins) Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net> Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
2015-05-18 22:11:12 +02:00
static uint64_t *setup_new_table(uint64_t desc, size_t xlat_size)
{
arm64: Decouple MMU functions from memranges The current arm64 MMU interface is difficult to use in pre-RAM environments. It is based on the memranges API which makes use of malloc(), and early stages usually don't have a heap. It is also built as a one-shot interface that requires all memory ranges to be laid out beforehand, which is a problem when existing areas need to change (e.g. after initializing DRAM). The long-term goal of this patch is to completely switch to a configure-as-you-go interface based on the mmu_config_range() function, similar to what ARM32 does. As a first step this feature is added side-by-side to the existing interface so that existing SoC implementations continue to work and can be slowly ported over one by one. Like the ARM32 version it does not garbage collect page tables that become unused, so repeated mapping at different granularities will exhaust the available table space (this is presumed to be a reasonable limitation for a firmware environment and keeps the code much simpler). Also do some cleanup, align comments between coreboot and libpayload for easier diffing, and change all error cases to assert()s. Right now the code just propagates error codes up the stack until it eventually reaches a function that doesn't check them anymore. MMU configuration errors (essentially just misaligned requests and running out of table space) should always be compile-time programming errors, so failing hard and fast seems like the best way to deal with them. BRANCH=None BUG=None TEST=Compile-tested rush_ryu. Booted on Oak and hacked MMU init to use mmu_config_range() insted of memranges. Confirmed that CRCs over all page tables before and after the change are equal. Change-Id: I93585b44a277c1d96d31ee9c3dd2522b5e10085b Signed-off-by: Patrick Georgi <pgeorgi@chromium.org> Original-Commit-Id: f10fcba107aba1f3ea239471cb5a4f9239809539 Original-Change-Id: I6a2a11e3b94e6ae9e1553871f0cccd3b556b3e65 Original-Signed-off-by: Julius Werner <jwerner@chromium.org> Original-Reviewed-on: https://chromium-review.googlesource.com/271991 Original-Reviewed-by: Aaron Durbin <adurbin@chromium.org> Reviewed-on: http://review.coreboot.org/10304 Tested-by: build bot (Jenkins) Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net> Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
2015-05-18 22:11:12 +02:00
uint64_t *new, *entry;
arm64: Decouple MMU functions from memranges The current arm64 MMU interface is difficult to use in pre-RAM environments. It is based on the memranges API which makes use of malloc(), and early stages usually don't have a heap. It is also built as a one-shot interface that requires all memory ranges to be laid out beforehand, which is a problem when existing areas need to change (e.g. after initializing DRAM). The long-term goal of this patch is to completely switch to a configure-as-you-go interface based on the mmu_config_range() function, similar to what ARM32 does. As a first step this feature is added side-by-side to the existing interface so that existing SoC implementations continue to work and can be slowly ported over one by one. Like the ARM32 version it does not garbage collect page tables that become unused, so repeated mapping at different granularities will exhaust the available table space (this is presumed to be a reasonable limitation for a firmware environment and keeps the code much simpler). Also do some cleanup, align comments between coreboot and libpayload for easier diffing, and change all error cases to assert()s. Right now the code just propagates error codes up the stack until it eventually reaches a function that doesn't check them anymore. MMU configuration errors (essentially just misaligned requests and running out of table space) should always be compile-time programming errors, so failing hard and fast seems like the best way to deal with them. BRANCH=None BUG=None TEST=Compile-tested rush_ryu. Booted on Oak and hacked MMU init to use mmu_config_range() insted of memranges. Confirmed that CRCs over all page tables before and after the change are equal. Change-Id: I93585b44a277c1d96d31ee9c3dd2522b5e10085b Signed-off-by: Patrick Georgi <pgeorgi@chromium.org> Original-Commit-Id: f10fcba107aba1f3ea239471cb5a4f9239809539 Original-Change-Id: I6a2a11e3b94e6ae9e1553871f0cccd3b556b3e65 Original-Signed-off-by: Julius Werner <jwerner@chromium.org> Original-Reviewed-on: https://chromium-review.googlesource.com/271991 Original-Reviewed-by: Aaron Durbin <adurbin@chromium.org> Reviewed-on: http://review.coreboot.org/10304 Tested-by: build bot (Jenkins) Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net> Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
2015-05-18 22:11:12 +02:00
assert(free_idx < max_tables);
new = (uint64_t*)((unsigned char *)xlat_addr + free_idx * GRANULE_SIZE);
free_idx++;
arm64: Decouple MMU functions from memranges The current arm64 MMU interface is difficult to use in pre-RAM environments. It is based on the memranges API which makes use of malloc(), and early stages usually don't have a heap. It is also built as a one-shot interface that requires all memory ranges to be laid out beforehand, which is a problem when existing areas need to change (e.g. after initializing DRAM). The long-term goal of this patch is to completely switch to a configure-as-you-go interface based on the mmu_config_range() function, similar to what ARM32 does. As a first step this feature is added side-by-side to the existing interface so that existing SoC implementations continue to work and can be slowly ported over one by one. Like the ARM32 version it does not garbage collect page tables that become unused, so repeated mapping at different granularities will exhaust the available table space (this is presumed to be a reasonable limitation for a firmware environment and keeps the code much simpler). Also do some cleanup, align comments between coreboot and libpayload for easier diffing, and change all error cases to assert()s. Right now the code just propagates error codes up the stack until it eventually reaches a function that doesn't check them anymore. MMU configuration errors (essentially just misaligned requests and running out of table space) should always be compile-time programming errors, so failing hard and fast seems like the best way to deal with them. BRANCH=None BUG=None TEST=Compile-tested rush_ryu. Booted on Oak and hacked MMU init to use mmu_config_range() insted of memranges. Confirmed that CRCs over all page tables before and after the change are equal. Change-Id: I93585b44a277c1d96d31ee9c3dd2522b5e10085b Signed-off-by: Patrick Georgi <pgeorgi@chromium.org> Original-Commit-Id: f10fcba107aba1f3ea239471cb5a4f9239809539 Original-Change-Id: I6a2a11e3b94e6ae9e1553871f0cccd3b556b3e65 Original-Signed-off-by: Julius Werner <jwerner@chromium.org> Original-Reviewed-on: https://chromium-review.googlesource.com/271991 Original-Reviewed-by: Aaron Durbin <adurbin@chromium.org> Reviewed-on: http://review.coreboot.org/10304 Tested-by: build bot (Jenkins) Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net> Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
2015-05-18 22:11:12 +02:00
if (!desc) {
memset(new, 0, GRANULE_SIZE);
} else {
/* Can reuse old parent entry, but may need to adjust type. */
if (xlat_size == L3_XLAT_SIZE)
desc |= PAGE_DESC;
for (entry = new; (u8 *)entry < (u8 *)new + GRANULE_SIZE;
entry++, desc += xlat_size)
*entry = desc;
}
return new;
}
arm64: Decouple MMU functions from memranges The current arm64 MMU interface is difficult to use in pre-RAM environments. It is based on the memranges API which makes use of malloc(), and early stages usually don't have a heap. It is also built as a one-shot interface that requires all memory ranges to be laid out beforehand, which is a problem when existing areas need to change (e.g. after initializing DRAM). The long-term goal of this patch is to completely switch to a configure-as-you-go interface based on the mmu_config_range() function, similar to what ARM32 does. As a first step this feature is added side-by-side to the existing interface so that existing SoC implementations continue to work and can be slowly ported over one by one. Like the ARM32 version it does not garbage collect page tables that become unused, so repeated mapping at different granularities will exhaust the available table space (this is presumed to be a reasonable limitation for a firmware environment and keeps the code much simpler). Also do some cleanup, align comments between coreboot and libpayload for easier diffing, and change all error cases to assert()s. Right now the code just propagates error codes up the stack until it eventually reaches a function that doesn't check them anymore. MMU configuration errors (essentially just misaligned requests and running out of table space) should always be compile-time programming errors, so failing hard and fast seems like the best way to deal with them. BRANCH=None BUG=None TEST=Compile-tested rush_ryu. Booted on Oak and hacked MMU init to use mmu_config_range() insted of memranges. Confirmed that CRCs over all page tables before and after the change are equal. Change-Id: I93585b44a277c1d96d31ee9c3dd2522b5e10085b Signed-off-by: Patrick Georgi <pgeorgi@chromium.org> Original-Commit-Id: f10fcba107aba1f3ea239471cb5a4f9239809539 Original-Change-Id: I6a2a11e3b94e6ae9e1553871f0cccd3b556b3e65 Original-Signed-off-by: Julius Werner <jwerner@chromium.org> Original-Reviewed-on: https://chromium-review.googlesource.com/271991 Original-Reviewed-by: Aaron Durbin <adurbin@chromium.org> Reviewed-on: http://review.coreboot.org/10304 Tested-by: build bot (Jenkins) Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net> Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
2015-05-18 22:11:12 +02:00
/* Func : get_table_from_desc
* Desc : Get next level table address from table descriptor
*/
static uint64_t *get_table_from_desc(uint64_t desc)
{
uint64_t *ptr = (uint64_t*)(desc & XLAT_TABLE_MASK);
return ptr;
}
arm64: Decouple MMU functions from memranges The current arm64 MMU interface is difficult to use in pre-RAM environments. It is based on the memranges API which makes use of malloc(), and early stages usually don't have a heap. It is also built as a one-shot interface that requires all memory ranges to be laid out beforehand, which is a problem when existing areas need to change (e.g. after initializing DRAM). The long-term goal of this patch is to completely switch to a configure-as-you-go interface based on the mmu_config_range() function, similar to what ARM32 does. As a first step this feature is added side-by-side to the existing interface so that existing SoC implementations continue to work and can be slowly ported over one by one. Like the ARM32 version it does not garbage collect page tables that become unused, so repeated mapping at different granularities will exhaust the available table space (this is presumed to be a reasonable limitation for a firmware environment and keeps the code much simpler). Also do some cleanup, align comments between coreboot and libpayload for easier diffing, and change all error cases to assert()s. Right now the code just propagates error codes up the stack until it eventually reaches a function that doesn't check them anymore. MMU configuration errors (essentially just misaligned requests and running out of table space) should always be compile-time programming errors, so failing hard and fast seems like the best way to deal with them. BRANCH=None BUG=None TEST=Compile-tested rush_ryu. Booted on Oak and hacked MMU init to use mmu_config_range() insted of memranges. Confirmed that CRCs over all page tables before and after the change are equal. Change-Id: I93585b44a277c1d96d31ee9c3dd2522b5e10085b Signed-off-by: Patrick Georgi <pgeorgi@chromium.org> Original-Commit-Id: f10fcba107aba1f3ea239471cb5a4f9239809539 Original-Change-Id: I6a2a11e3b94e6ae9e1553871f0cccd3b556b3e65 Original-Signed-off-by: Julius Werner <jwerner@chromium.org> Original-Reviewed-on: https://chromium-review.googlesource.com/271991 Original-Reviewed-by: Aaron Durbin <adurbin@chromium.org> Reviewed-on: http://review.coreboot.org/10304 Tested-by: build bot (Jenkins) Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net> Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
2015-05-18 22:11:12 +02:00
/* Func: get_next_level_table
* Desc: Check if the table entry is a valid descriptor. If not, initialize new
* table, update the entry and return the table addr. If valid, return the addr.
*/
arm64: Decouple MMU functions from memranges The current arm64 MMU interface is difficult to use in pre-RAM environments. It is based on the memranges API which makes use of malloc(), and early stages usually don't have a heap. It is also built as a one-shot interface that requires all memory ranges to be laid out beforehand, which is a problem when existing areas need to change (e.g. after initializing DRAM). The long-term goal of this patch is to completely switch to a configure-as-you-go interface based on the mmu_config_range() function, similar to what ARM32 does. As a first step this feature is added side-by-side to the existing interface so that existing SoC implementations continue to work and can be slowly ported over one by one. Like the ARM32 version it does not garbage collect page tables that become unused, so repeated mapping at different granularities will exhaust the available table space (this is presumed to be a reasonable limitation for a firmware environment and keeps the code much simpler). Also do some cleanup, align comments between coreboot and libpayload for easier diffing, and change all error cases to assert()s. Right now the code just propagates error codes up the stack until it eventually reaches a function that doesn't check them anymore. MMU configuration errors (essentially just misaligned requests and running out of table space) should always be compile-time programming errors, so failing hard and fast seems like the best way to deal with them. BRANCH=None BUG=None TEST=Compile-tested rush_ryu. Booted on Oak and hacked MMU init to use mmu_config_range() insted of memranges. Confirmed that CRCs over all page tables before and after the change are equal. Change-Id: I93585b44a277c1d96d31ee9c3dd2522b5e10085b Signed-off-by: Patrick Georgi <pgeorgi@chromium.org> Original-Commit-Id: f10fcba107aba1f3ea239471cb5a4f9239809539 Original-Change-Id: I6a2a11e3b94e6ae9e1553871f0cccd3b556b3e65 Original-Signed-off-by: Julius Werner <jwerner@chromium.org> Original-Reviewed-on: https://chromium-review.googlesource.com/271991 Original-Reviewed-by: Aaron Durbin <adurbin@chromium.org> Reviewed-on: http://review.coreboot.org/10304 Tested-by: build bot (Jenkins) Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net> Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
2015-05-18 22:11:12 +02:00
static uint64_t *get_next_level_table(uint64_t *ptr, size_t xlat_size)
{
uint64_t desc = *ptr;
if (!table_desc_valid(desc)) {
arm64: Decouple MMU functions from memranges The current arm64 MMU interface is difficult to use in pre-RAM environments. It is based on the memranges API which makes use of malloc(), and early stages usually don't have a heap. It is also built as a one-shot interface that requires all memory ranges to be laid out beforehand, which is a problem when existing areas need to change (e.g. after initializing DRAM). The long-term goal of this patch is to completely switch to a configure-as-you-go interface based on the mmu_config_range() function, similar to what ARM32 does. As a first step this feature is added side-by-side to the existing interface so that existing SoC implementations continue to work and can be slowly ported over one by one. Like the ARM32 version it does not garbage collect page tables that become unused, so repeated mapping at different granularities will exhaust the available table space (this is presumed to be a reasonable limitation for a firmware environment and keeps the code much simpler). Also do some cleanup, align comments between coreboot and libpayload for easier diffing, and change all error cases to assert()s. Right now the code just propagates error codes up the stack until it eventually reaches a function that doesn't check them anymore. MMU configuration errors (essentially just misaligned requests and running out of table space) should always be compile-time programming errors, so failing hard and fast seems like the best way to deal with them. BRANCH=None BUG=None TEST=Compile-tested rush_ryu. Booted on Oak and hacked MMU init to use mmu_config_range() insted of memranges. Confirmed that CRCs over all page tables before and after the change are equal. Change-Id: I93585b44a277c1d96d31ee9c3dd2522b5e10085b Signed-off-by: Patrick Georgi <pgeorgi@chromium.org> Original-Commit-Id: f10fcba107aba1f3ea239471cb5a4f9239809539 Original-Change-Id: I6a2a11e3b94e6ae9e1553871f0cccd3b556b3e65 Original-Signed-off-by: Julius Werner <jwerner@chromium.org> Original-Reviewed-on: https://chromium-review.googlesource.com/271991 Original-Reviewed-by: Aaron Durbin <adurbin@chromium.org> Reviewed-on: http://review.coreboot.org/10304 Tested-by: build bot (Jenkins) Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net> Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
2015-05-18 22:11:12 +02:00
uint64_t *new_table = setup_new_table(desc, xlat_size);
desc = ((uint64_t)new_table) | TABLE_DESC;
*ptr = desc;
}
return get_table_from_desc(desc);
}
arm64: Decouple MMU functions from memranges The current arm64 MMU interface is difficult to use in pre-RAM environments. It is based on the memranges API which makes use of malloc(), and early stages usually don't have a heap. It is also built as a one-shot interface that requires all memory ranges to be laid out beforehand, which is a problem when existing areas need to change (e.g. after initializing DRAM). The long-term goal of this patch is to completely switch to a configure-as-you-go interface based on the mmu_config_range() function, similar to what ARM32 does. As a first step this feature is added side-by-side to the existing interface so that existing SoC implementations continue to work and can be slowly ported over one by one. Like the ARM32 version it does not garbage collect page tables that become unused, so repeated mapping at different granularities will exhaust the available table space (this is presumed to be a reasonable limitation for a firmware environment and keeps the code much simpler). Also do some cleanup, align comments between coreboot and libpayload for easier diffing, and change all error cases to assert()s. Right now the code just propagates error codes up the stack until it eventually reaches a function that doesn't check them anymore. MMU configuration errors (essentially just misaligned requests and running out of table space) should always be compile-time programming errors, so failing hard and fast seems like the best way to deal with them. BRANCH=None BUG=None TEST=Compile-tested rush_ryu. Booted on Oak and hacked MMU init to use mmu_config_range() insted of memranges. Confirmed that CRCs over all page tables before and after the change are equal. Change-Id: I93585b44a277c1d96d31ee9c3dd2522b5e10085b Signed-off-by: Patrick Georgi <pgeorgi@chromium.org> Original-Commit-Id: f10fcba107aba1f3ea239471cb5a4f9239809539 Original-Change-Id: I6a2a11e3b94e6ae9e1553871f0cccd3b556b3e65 Original-Signed-off-by: Julius Werner <jwerner@chromium.org> Original-Reviewed-on: https://chromium-review.googlesource.com/271991 Original-Reviewed-by: Aaron Durbin <adurbin@chromium.org> Reviewed-on: http://review.coreboot.org/10304 Tested-by: build bot (Jenkins) Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net> Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
2015-05-18 22:11:12 +02:00
/* Func : init_xlat_table
* Desc : Given a base address and size, it identifies the indices within
* different level XLAT tables which map the given base addr. Similar to table
* walk, except that all invalid entries during the walk are updated
* accordingly. On success, it returns the size of the block/page addressed by
* the final table.
*/
static uint64_t init_xlat_table(uint64_t base_addr,
uint64_t size,
uint64_t tag)
{
uint64_t l0_index = (base_addr & L0_ADDR_MASK) >> L0_ADDR_SHIFT;
arm64: Decouple MMU functions from memranges The current arm64 MMU interface is difficult to use in pre-RAM environments. It is based on the memranges API which makes use of malloc(), and early stages usually don't have a heap. It is also built as a one-shot interface that requires all memory ranges to be laid out beforehand, which is a problem when existing areas need to change (e.g. after initializing DRAM). The long-term goal of this patch is to completely switch to a configure-as-you-go interface based on the mmu_config_range() function, similar to what ARM32 does. As a first step this feature is added side-by-side to the existing interface so that existing SoC implementations continue to work and can be slowly ported over one by one. Like the ARM32 version it does not garbage collect page tables that become unused, so repeated mapping at different granularities will exhaust the available table space (this is presumed to be a reasonable limitation for a firmware environment and keeps the code much simpler). Also do some cleanup, align comments between coreboot and libpayload for easier diffing, and change all error cases to assert()s. Right now the code just propagates error codes up the stack until it eventually reaches a function that doesn't check them anymore. MMU configuration errors (essentially just misaligned requests and running out of table space) should always be compile-time programming errors, so failing hard and fast seems like the best way to deal with them. BRANCH=None BUG=None TEST=Compile-tested rush_ryu. Booted on Oak and hacked MMU init to use mmu_config_range() insted of memranges. Confirmed that CRCs over all page tables before and after the change are equal. Change-Id: I93585b44a277c1d96d31ee9c3dd2522b5e10085b Signed-off-by: Patrick Georgi <pgeorgi@chromium.org> Original-Commit-Id: f10fcba107aba1f3ea239471cb5a4f9239809539 Original-Change-Id: I6a2a11e3b94e6ae9e1553871f0cccd3b556b3e65 Original-Signed-off-by: Julius Werner <jwerner@chromium.org> Original-Reviewed-on: https://chromium-review.googlesource.com/271991 Original-Reviewed-by: Aaron Durbin <adurbin@chromium.org> Reviewed-on: http://review.coreboot.org/10304 Tested-by: build bot (Jenkins) Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net> Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
2015-05-18 22:11:12 +02:00
uint64_t l1_index = (base_addr & L1_ADDR_MASK) >> L1_ADDR_SHIFT;
uint64_t l2_index = (base_addr & L2_ADDR_MASK) >> L2_ADDR_SHIFT;
uint64_t l3_index = (base_addr & L3_ADDR_MASK) >> L3_ADDR_SHIFT;
uint64_t *table = xlat_addr;
uint64_t desc;
uint64_t attr = get_block_attr(tag);
/* L0 entry stores a table descriptor (doesn't support blocks) */
table = get_next_level_table(&table[l0_index], L1_XLAT_SIZE);
/* L1 table lookup */
if ((size >= L1_XLAT_SIZE) &&
IS_ALIGNED(base_addr, (1UL << L1_ADDR_SHIFT))) {
/* If block address is aligned and size is greater than
* or equal to size addressed by each L1 entry, we can
* directly store a block desc */
desc = base_addr | BLOCK_DESC | attr;
table[l1_index] = desc;
/* L2 lookup is not required */
return L1_XLAT_SIZE;
}
/* L1 entry stores a table descriptor */
table = get_next_level_table(&table[l1_index], L2_XLAT_SIZE);
/* L2 table lookup */
if ((size >= L2_XLAT_SIZE) &&
IS_ALIGNED(base_addr, (1UL << L2_ADDR_SHIFT))) {
arm64: Decouple MMU functions from memranges The current arm64 MMU interface is difficult to use in pre-RAM environments. It is based on the memranges API which makes use of malloc(), and early stages usually don't have a heap. It is also built as a one-shot interface that requires all memory ranges to be laid out beforehand, which is a problem when existing areas need to change (e.g. after initializing DRAM). The long-term goal of this patch is to completely switch to a configure-as-you-go interface based on the mmu_config_range() function, similar to what ARM32 does. As a first step this feature is added side-by-side to the existing interface so that existing SoC implementations continue to work and can be slowly ported over one by one. Like the ARM32 version it does not garbage collect page tables that become unused, so repeated mapping at different granularities will exhaust the available table space (this is presumed to be a reasonable limitation for a firmware environment and keeps the code much simpler). Also do some cleanup, align comments between coreboot and libpayload for easier diffing, and change all error cases to assert()s. Right now the code just propagates error codes up the stack until it eventually reaches a function that doesn't check them anymore. MMU configuration errors (essentially just misaligned requests and running out of table space) should always be compile-time programming errors, so failing hard and fast seems like the best way to deal with them. BRANCH=None BUG=None TEST=Compile-tested rush_ryu. Booted on Oak and hacked MMU init to use mmu_config_range() insted of memranges. Confirmed that CRCs over all page tables before and after the change are equal. Change-Id: I93585b44a277c1d96d31ee9c3dd2522b5e10085b Signed-off-by: Patrick Georgi <pgeorgi@chromium.org> Original-Commit-Id: f10fcba107aba1f3ea239471cb5a4f9239809539 Original-Change-Id: I6a2a11e3b94e6ae9e1553871f0cccd3b556b3e65 Original-Signed-off-by: Julius Werner <jwerner@chromium.org> Original-Reviewed-on: https://chromium-review.googlesource.com/271991 Original-Reviewed-by: Aaron Durbin <adurbin@chromium.org> Reviewed-on: http://review.coreboot.org/10304 Tested-by: build bot (Jenkins) Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net> Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
2015-05-18 22:11:12 +02:00
/* If block address is aligned and size is greater than
* or equal to size addressed by each L2 entry, we can
* directly store a block desc */
desc = base_addr | BLOCK_DESC | attr;
table[l2_index] = desc;
/* L3 lookup is not required */
return L2_XLAT_SIZE;
}
arm64: Decouple MMU functions from memranges The current arm64 MMU interface is difficult to use in pre-RAM environments. It is based on the memranges API which makes use of malloc(), and early stages usually don't have a heap. It is also built as a one-shot interface that requires all memory ranges to be laid out beforehand, which is a problem when existing areas need to change (e.g. after initializing DRAM). The long-term goal of this patch is to completely switch to a configure-as-you-go interface based on the mmu_config_range() function, similar to what ARM32 does. As a first step this feature is added side-by-side to the existing interface so that existing SoC implementations continue to work and can be slowly ported over one by one. Like the ARM32 version it does not garbage collect page tables that become unused, so repeated mapping at different granularities will exhaust the available table space (this is presumed to be a reasonable limitation for a firmware environment and keeps the code much simpler). Also do some cleanup, align comments between coreboot and libpayload for easier diffing, and change all error cases to assert()s. Right now the code just propagates error codes up the stack until it eventually reaches a function that doesn't check them anymore. MMU configuration errors (essentially just misaligned requests and running out of table space) should always be compile-time programming errors, so failing hard and fast seems like the best way to deal with them. BRANCH=None BUG=None TEST=Compile-tested rush_ryu. Booted on Oak and hacked MMU init to use mmu_config_range() insted of memranges. Confirmed that CRCs over all page tables before and after the change are equal. Change-Id: I93585b44a277c1d96d31ee9c3dd2522b5e10085b Signed-off-by: Patrick Georgi <pgeorgi@chromium.org> Original-Commit-Id: f10fcba107aba1f3ea239471cb5a4f9239809539 Original-Change-Id: I6a2a11e3b94e6ae9e1553871f0cccd3b556b3e65 Original-Signed-off-by: Julius Werner <jwerner@chromium.org> Original-Reviewed-on: https://chromium-review.googlesource.com/271991 Original-Reviewed-by: Aaron Durbin <adurbin@chromium.org> Reviewed-on: http://review.coreboot.org/10304 Tested-by: build bot (Jenkins) Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net> Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
2015-05-18 22:11:12 +02:00
/* L2 entry stores a table descriptor */
table = get_next_level_table(&table[l2_index], L3_XLAT_SIZE);
/* L3 table lookup */
desc = base_addr | PAGE_DESC | attr;
table[l3_index] = desc;
return L3_XLAT_SIZE;
}
arm64: Decouple MMU functions from memranges The current arm64 MMU interface is difficult to use in pre-RAM environments. It is based on the memranges API which makes use of malloc(), and early stages usually don't have a heap. It is also built as a one-shot interface that requires all memory ranges to be laid out beforehand, which is a problem when existing areas need to change (e.g. after initializing DRAM). The long-term goal of this patch is to completely switch to a configure-as-you-go interface based on the mmu_config_range() function, similar to what ARM32 does. As a first step this feature is added side-by-side to the existing interface so that existing SoC implementations continue to work and can be slowly ported over one by one. Like the ARM32 version it does not garbage collect page tables that become unused, so repeated mapping at different granularities will exhaust the available table space (this is presumed to be a reasonable limitation for a firmware environment and keeps the code much simpler). Also do some cleanup, align comments between coreboot and libpayload for easier diffing, and change all error cases to assert()s. Right now the code just propagates error codes up the stack until it eventually reaches a function that doesn't check them anymore. MMU configuration errors (essentially just misaligned requests and running out of table space) should always be compile-time programming errors, so failing hard and fast seems like the best way to deal with them. BRANCH=None BUG=None TEST=Compile-tested rush_ryu. Booted on Oak and hacked MMU init to use mmu_config_range() insted of memranges. Confirmed that CRCs over all page tables before and after the change are equal. Change-Id: I93585b44a277c1d96d31ee9c3dd2522b5e10085b Signed-off-by: Patrick Georgi <pgeorgi@chromium.org> Original-Commit-Id: f10fcba107aba1f3ea239471cb5a4f9239809539 Original-Change-Id: I6a2a11e3b94e6ae9e1553871f0cccd3b556b3e65 Original-Signed-off-by: Julius Werner <jwerner@chromium.org> Original-Reviewed-on: https://chromium-review.googlesource.com/271991 Original-Reviewed-by: Aaron Durbin <adurbin@chromium.org> Reviewed-on: http://review.coreboot.org/10304 Tested-by: build bot (Jenkins) Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net> Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
2015-05-18 22:11:12 +02:00
/* Func : sanity_check
* Desc : Check address/size alignment of a table or page.
*/
arm64: Decouple MMU functions from memranges The current arm64 MMU interface is difficult to use in pre-RAM environments. It is based on the memranges API which makes use of malloc(), and early stages usually don't have a heap. It is also built as a one-shot interface that requires all memory ranges to be laid out beforehand, which is a problem when existing areas need to change (e.g. after initializing DRAM). The long-term goal of this patch is to completely switch to a configure-as-you-go interface based on the mmu_config_range() function, similar to what ARM32 does. As a first step this feature is added side-by-side to the existing interface so that existing SoC implementations continue to work and can be slowly ported over one by one. Like the ARM32 version it does not garbage collect page tables that become unused, so repeated mapping at different granularities will exhaust the available table space (this is presumed to be a reasonable limitation for a firmware environment and keeps the code much simpler). Also do some cleanup, align comments between coreboot and libpayload for easier diffing, and change all error cases to assert()s. Right now the code just propagates error codes up the stack until it eventually reaches a function that doesn't check them anymore. MMU configuration errors (essentially just misaligned requests and running out of table space) should always be compile-time programming errors, so failing hard and fast seems like the best way to deal with them. BRANCH=None BUG=None TEST=Compile-tested rush_ryu. Booted on Oak and hacked MMU init to use mmu_config_range() insted of memranges. Confirmed that CRCs over all page tables before and after the change are equal. Change-Id: I93585b44a277c1d96d31ee9c3dd2522b5e10085b Signed-off-by: Patrick Georgi <pgeorgi@chromium.org> Original-Commit-Id: f10fcba107aba1f3ea239471cb5a4f9239809539 Original-Change-Id: I6a2a11e3b94e6ae9e1553871f0cccd3b556b3e65 Original-Signed-off-by: Julius Werner <jwerner@chromium.org> Original-Reviewed-on: https://chromium-review.googlesource.com/271991 Original-Reviewed-by: Aaron Durbin <adurbin@chromium.org> Reviewed-on: http://review.coreboot.org/10304 Tested-by: build bot (Jenkins) Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net> Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
2015-05-18 22:11:12 +02:00
static void sanity_check(uint64_t addr, uint64_t size)
{
arm64: Decouple MMU functions from memranges The current arm64 MMU interface is difficult to use in pre-RAM environments. It is based on the memranges API which makes use of malloc(), and early stages usually don't have a heap. It is also built as a one-shot interface that requires all memory ranges to be laid out beforehand, which is a problem when existing areas need to change (e.g. after initializing DRAM). The long-term goal of this patch is to completely switch to a configure-as-you-go interface based on the mmu_config_range() function, similar to what ARM32 does. As a first step this feature is added side-by-side to the existing interface so that existing SoC implementations continue to work and can be slowly ported over one by one. Like the ARM32 version it does not garbage collect page tables that become unused, so repeated mapping at different granularities will exhaust the available table space (this is presumed to be a reasonable limitation for a firmware environment and keeps the code much simpler). Also do some cleanup, align comments between coreboot and libpayload for easier diffing, and change all error cases to assert()s. Right now the code just propagates error codes up the stack until it eventually reaches a function that doesn't check them anymore. MMU configuration errors (essentially just misaligned requests and running out of table space) should always be compile-time programming errors, so failing hard and fast seems like the best way to deal with them. BRANCH=None BUG=None TEST=Compile-tested rush_ryu. Booted on Oak and hacked MMU init to use mmu_config_range() insted of memranges. Confirmed that CRCs over all page tables before and after the change are equal. Change-Id: I93585b44a277c1d96d31ee9c3dd2522b5e10085b Signed-off-by: Patrick Georgi <pgeorgi@chromium.org> Original-Commit-Id: f10fcba107aba1f3ea239471cb5a4f9239809539 Original-Change-Id: I6a2a11e3b94e6ae9e1553871f0cccd3b556b3e65 Original-Signed-off-by: Julius Werner <jwerner@chromium.org> Original-Reviewed-on: https://chromium-review.googlesource.com/271991 Original-Reviewed-by: Aaron Durbin <adurbin@chromium.org> Reviewed-on: http://review.coreboot.org/10304 Tested-by: build bot (Jenkins) Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net> Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
2015-05-18 22:11:12 +02:00
assert(!(addr & GRANULE_SIZE_MASK) &&
!(size & GRANULE_SIZE_MASK) &&
(addr + size < (1UL << BITS_PER_VA)) &&
arm64: Decouple MMU functions from memranges The current arm64 MMU interface is difficult to use in pre-RAM environments. It is based on the memranges API which makes use of malloc(), and early stages usually don't have a heap. It is also built as a one-shot interface that requires all memory ranges to be laid out beforehand, which is a problem when existing areas need to change (e.g. after initializing DRAM). The long-term goal of this patch is to completely switch to a configure-as-you-go interface based on the mmu_config_range() function, similar to what ARM32 does. As a first step this feature is added side-by-side to the existing interface so that existing SoC implementations continue to work and can be slowly ported over one by one. Like the ARM32 version it does not garbage collect page tables that become unused, so repeated mapping at different granularities will exhaust the available table space (this is presumed to be a reasonable limitation for a firmware environment and keeps the code much simpler). Also do some cleanup, align comments between coreboot and libpayload for easier diffing, and change all error cases to assert()s. Right now the code just propagates error codes up the stack until it eventually reaches a function that doesn't check them anymore. MMU configuration errors (essentially just misaligned requests and running out of table space) should always be compile-time programming errors, so failing hard and fast seems like the best way to deal with them. BRANCH=None BUG=None TEST=Compile-tested rush_ryu. Booted on Oak and hacked MMU init to use mmu_config_range() insted of memranges. Confirmed that CRCs over all page tables before and after the change are equal. Change-Id: I93585b44a277c1d96d31ee9c3dd2522b5e10085b Signed-off-by: Patrick Georgi <pgeorgi@chromium.org> Original-Commit-Id: f10fcba107aba1f3ea239471cb5a4f9239809539 Original-Change-Id: I6a2a11e3b94e6ae9e1553871f0cccd3b556b3e65 Original-Signed-off-by: Julius Werner <jwerner@chromium.org> Original-Reviewed-on: https://chromium-review.googlesource.com/271991 Original-Reviewed-by: Aaron Durbin <adurbin@chromium.org> Reviewed-on: http://review.coreboot.org/10304 Tested-by: build bot (Jenkins) Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net> Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
2015-05-18 22:11:12 +02:00
size >= GRANULE_SIZE);
}
arm64: Decouple MMU functions from memranges The current arm64 MMU interface is difficult to use in pre-RAM environments. It is based on the memranges API which makes use of malloc(), and early stages usually don't have a heap. It is also built as a one-shot interface that requires all memory ranges to be laid out beforehand, which is a problem when existing areas need to change (e.g. after initializing DRAM). The long-term goal of this patch is to completely switch to a configure-as-you-go interface based on the mmu_config_range() function, similar to what ARM32 does. As a first step this feature is added side-by-side to the existing interface so that existing SoC implementations continue to work and can be slowly ported over one by one. Like the ARM32 version it does not garbage collect page tables that become unused, so repeated mapping at different granularities will exhaust the available table space (this is presumed to be a reasonable limitation for a firmware environment and keeps the code much simpler). Also do some cleanup, align comments between coreboot and libpayload for easier diffing, and change all error cases to assert()s. Right now the code just propagates error codes up the stack until it eventually reaches a function that doesn't check them anymore. MMU configuration errors (essentially just misaligned requests and running out of table space) should always be compile-time programming errors, so failing hard and fast seems like the best way to deal with them. BRANCH=None BUG=None TEST=Compile-tested rush_ryu. Booted on Oak and hacked MMU init to use mmu_config_range() insted of memranges. Confirmed that CRCs over all page tables before and after the change are equal. Change-Id: I93585b44a277c1d96d31ee9c3dd2522b5e10085b Signed-off-by: Patrick Georgi <pgeorgi@chromium.org> Original-Commit-Id: f10fcba107aba1f3ea239471cb5a4f9239809539 Original-Change-Id: I6a2a11e3b94e6ae9e1553871f0cccd3b556b3e65 Original-Signed-off-by: Julius Werner <jwerner@chromium.org> Original-Reviewed-on: https://chromium-review.googlesource.com/271991 Original-Reviewed-by: Aaron Durbin <adurbin@chromium.org> Reviewed-on: http://review.coreboot.org/10304 Tested-by: build bot (Jenkins) Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net> Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
2015-05-18 22:11:12 +02:00
/* Func : mmu_config_range
* Desc : This function repeatedly calls init_xlat_table with the base
* address. Based on size returned from init_xlat_table, base_addr is updated
* and subsequent calls are made for initializing the xlat table until the whole
* region is initialized.
*/
arm64: Decouple MMU functions from memranges The current arm64 MMU interface is difficult to use in pre-RAM environments. It is based on the memranges API which makes use of malloc(), and early stages usually don't have a heap. It is also built as a one-shot interface that requires all memory ranges to be laid out beforehand, which is a problem when existing areas need to change (e.g. after initializing DRAM). The long-term goal of this patch is to completely switch to a configure-as-you-go interface based on the mmu_config_range() function, similar to what ARM32 does. As a first step this feature is added side-by-side to the existing interface so that existing SoC implementations continue to work and can be slowly ported over one by one. Like the ARM32 version it does not garbage collect page tables that become unused, so repeated mapping at different granularities will exhaust the available table space (this is presumed to be a reasonable limitation for a firmware environment and keeps the code much simpler). Also do some cleanup, align comments between coreboot and libpayload for easier diffing, and change all error cases to assert()s. Right now the code just propagates error codes up the stack until it eventually reaches a function that doesn't check them anymore. MMU configuration errors (essentially just misaligned requests and running out of table space) should always be compile-time programming errors, so failing hard and fast seems like the best way to deal with them. BRANCH=None BUG=None TEST=Compile-tested rush_ryu. Booted on Oak and hacked MMU init to use mmu_config_range() insted of memranges. Confirmed that CRCs over all page tables before and after the change are equal. Change-Id: I93585b44a277c1d96d31ee9c3dd2522b5e10085b Signed-off-by: Patrick Georgi <pgeorgi@chromium.org> Original-Commit-Id: f10fcba107aba1f3ea239471cb5a4f9239809539 Original-Change-Id: I6a2a11e3b94e6ae9e1553871f0cccd3b556b3e65 Original-Signed-off-by: Julius Werner <jwerner@chromium.org> Original-Reviewed-on: https://chromium-review.googlesource.com/271991 Original-Reviewed-by: Aaron Durbin <adurbin@chromium.org> Reviewed-on: http://review.coreboot.org/10304 Tested-by: build bot (Jenkins) Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net> Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
2015-05-18 22:11:12 +02:00
void mmu_config_range(void *start, size_t size, uint64_t tag)
{
arm64: Decouple MMU functions from memranges The current arm64 MMU interface is difficult to use in pre-RAM environments. It is based on the memranges API which makes use of malloc(), and early stages usually don't have a heap. It is also built as a one-shot interface that requires all memory ranges to be laid out beforehand, which is a problem when existing areas need to change (e.g. after initializing DRAM). The long-term goal of this patch is to completely switch to a configure-as-you-go interface based on the mmu_config_range() function, similar to what ARM32 does. As a first step this feature is added side-by-side to the existing interface so that existing SoC implementations continue to work and can be slowly ported over one by one. Like the ARM32 version it does not garbage collect page tables that become unused, so repeated mapping at different granularities will exhaust the available table space (this is presumed to be a reasonable limitation for a firmware environment and keeps the code much simpler). Also do some cleanup, align comments between coreboot and libpayload for easier diffing, and change all error cases to assert()s. Right now the code just propagates error codes up the stack until it eventually reaches a function that doesn't check them anymore. MMU configuration errors (essentially just misaligned requests and running out of table space) should always be compile-time programming errors, so failing hard and fast seems like the best way to deal with them. BRANCH=None BUG=None TEST=Compile-tested rush_ryu. Booted on Oak and hacked MMU init to use mmu_config_range() insted of memranges. Confirmed that CRCs over all page tables before and after the change are equal. Change-Id: I93585b44a277c1d96d31ee9c3dd2522b5e10085b Signed-off-by: Patrick Georgi <pgeorgi@chromium.org> Original-Commit-Id: f10fcba107aba1f3ea239471cb5a4f9239809539 Original-Change-Id: I6a2a11e3b94e6ae9e1553871f0cccd3b556b3e65 Original-Signed-off-by: Julius Werner <jwerner@chromium.org> Original-Reviewed-on: https://chromium-review.googlesource.com/271991 Original-Reviewed-by: Aaron Durbin <adurbin@chromium.org> Reviewed-on: http://review.coreboot.org/10304 Tested-by: build bot (Jenkins) Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net> Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
2015-05-18 22:11:12 +02:00
uint64_t base_addr = (uintptr_t)start;
uint64_t temp_size = size;
arm64: Decouple MMU functions from memranges The current arm64 MMU interface is difficult to use in pre-RAM environments. It is based on the memranges API which makes use of malloc(), and early stages usually don't have a heap. It is also built as a one-shot interface that requires all memory ranges to be laid out beforehand, which is a problem when existing areas need to change (e.g. after initializing DRAM). The long-term goal of this patch is to completely switch to a configure-as-you-go interface based on the mmu_config_range() function, similar to what ARM32 does. As a first step this feature is added side-by-side to the existing interface so that existing SoC implementations continue to work and can be slowly ported over one by one. Like the ARM32 version it does not garbage collect page tables that become unused, so repeated mapping at different granularities will exhaust the available table space (this is presumed to be a reasonable limitation for a firmware environment and keeps the code much simpler). Also do some cleanup, align comments between coreboot and libpayload for easier diffing, and change all error cases to assert()s. Right now the code just propagates error codes up the stack until it eventually reaches a function that doesn't check them anymore. MMU configuration errors (essentially just misaligned requests and running out of table space) should always be compile-time programming errors, so failing hard and fast seems like the best way to deal with them. BRANCH=None BUG=None TEST=Compile-tested rush_ryu. Booted on Oak and hacked MMU init to use mmu_config_range() insted of memranges. Confirmed that CRCs over all page tables before and after the change are equal. Change-Id: I93585b44a277c1d96d31ee9c3dd2522b5e10085b Signed-off-by: Patrick Georgi <pgeorgi@chromium.org> Original-Commit-Id: f10fcba107aba1f3ea239471cb5a4f9239809539 Original-Change-Id: I6a2a11e3b94e6ae9e1553871f0cccd3b556b3e65 Original-Signed-off-by: Julius Werner <jwerner@chromium.org> Original-Reviewed-on: https://chromium-review.googlesource.com/271991 Original-Reviewed-by: Aaron Durbin <adurbin@chromium.org> Reviewed-on: http://review.coreboot.org/10304 Tested-by: build bot (Jenkins) Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net> Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
2015-05-18 22:11:12 +02:00
assert(tag < ARRAY_SIZE(tag_to_string));
printf("Libpayload: ARM64 MMU: Mapping address range [%p:%p) as %s\n",
start, start + size, tag_to_string[tag]);
sanity_check(base_addr, temp_size);
arm64: Decouple MMU functions from memranges The current arm64 MMU interface is difficult to use in pre-RAM environments. It is based on the memranges API which makes use of malloc(), and early stages usually don't have a heap. It is also built as a one-shot interface that requires all memory ranges to be laid out beforehand, which is a problem when existing areas need to change (e.g. after initializing DRAM). The long-term goal of this patch is to completely switch to a configure-as-you-go interface based on the mmu_config_range() function, similar to what ARM32 does. As a first step this feature is added side-by-side to the existing interface so that existing SoC implementations continue to work and can be slowly ported over one by one. Like the ARM32 version it does not garbage collect page tables that become unused, so repeated mapping at different granularities will exhaust the available table space (this is presumed to be a reasonable limitation for a firmware environment and keeps the code much simpler). Also do some cleanup, align comments between coreboot and libpayload for easier diffing, and change all error cases to assert()s. Right now the code just propagates error codes up the stack until it eventually reaches a function that doesn't check them anymore. MMU configuration errors (essentially just misaligned requests and running out of table space) should always be compile-time programming errors, so failing hard and fast seems like the best way to deal with them. BRANCH=None BUG=None TEST=Compile-tested rush_ryu. Booted on Oak and hacked MMU init to use mmu_config_range() insted of memranges. Confirmed that CRCs over all page tables before and after the change are equal. Change-Id: I93585b44a277c1d96d31ee9c3dd2522b5e10085b Signed-off-by: Patrick Georgi <pgeorgi@chromium.org> Original-Commit-Id: f10fcba107aba1f3ea239471cb5a4f9239809539 Original-Change-Id: I6a2a11e3b94e6ae9e1553871f0cccd3b556b3e65 Original-Signed-off-by: Julius Werner <jwerner@chromium.org> Original-Reviewed-on: https://chromium-review.googlesource.com/271991 Original-Reviewed-by: Aaron Durbin <adurbin@chromium.org> Reviewed-on: http://review.coreboot.org/10304 Tested-by: build bot (Jenkins) Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net> Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
2015-05-18 22:11:12 +02:00
while (temp_size)
temp_size -= init_xlat_table(base_addr + (size - temp_size),
temp_size, tag);
arm64: Decouple MMU functions from memranges The current arm64 MMU interface is difficult to use in pre-RAM environments. It is based on the memranges API which makes use of malloc(), and early stages usually don't have a heap. It is also built as a one-shot interface that requires all memory ranges to be laid out beforehand, which is a problem when existing areas need to change (e.g. after initializing DRAM). The long-term goal of this patch is to completely switch to a configure-as-you-go interface based on the mmu_config_range() function, similar to what ARM32 does. As a first step this feature is added side-by-side to the existing interface so that existing SoC implementations continue to work and can be slowly ported over one by one. Like the ARM32 version it does not garbage collect page tables that become unused, so repeated mapping at different granularities will exhaust the available table space (this is presumed to be a reasonable limitation for a firmware environment and keeps the code much simpler). Also do some cleanup, align comments between coreboot and libpayload for easier diffing, and change all error cases to assert()s. Right now the code just propagates error codes up the stack until it eventually reaches a function that doesn't check them anymore. MMU configuration errors (essentially just misaligned requests and running out of table space) should always be compile-time programming errors, so failing hard and fast seems like the best way to deal with them. BRANCH=None BUG=None TEST=Compile-tested rush_ryu. Booted on Oak and hacked MMU init to use mmu_config_range() insted of memranges. Confirmed that CRCs over all page tables before and after the change are equal. Change-Id: I93585b44a277c1d96d31ee9c3dd2522b5e10085b Signed-off-by: Patrick Georgi <pgeorgi@chromium.org> Original-Commit-Id: f10fcba107aba1f3ea239471cb5a4f9239809539 Original-Change-Id: I6a2a11e3b94e6ae9e1553871f0cccd3b556b3e65 Original-Signed-off-by: Julius Werner <jwerner@chromium.org> Original-Reviewed-on: https://chromium-review.googlesource.com/271991 Original-Reviewed-by: Aaron Durbin <adurbin@chromium.org> Reviewed-on: http://review.coreboot.org/10304 Tested-by: build bot (Jenkins) Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net> Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
2015-05-18 22:11:12 +02:00
/* ARMv8 MMUs snoop L1 data cache, no need to flush it. */
dsb();
tlbiall_el2();
arm64: Decouple MMU functions from memranges The current arm64 MMU interface is difficult to use in pre-RAM environments. It is based on the memranges API which makes use of malloc(), and early stages usually don't have a heap. It is also built as a one-shot interface that requires all memory ranges to be laid out beforehand, which is a problem when existing areas need to change (e.g. after initializing DRAM). The long-term goal of this patch is to completely switch to a configure-as-you-go interface based on the mmu_config_range() function, similar to what ARM32 does. As a first step this feature is added side-by-side to the existing interface so that existing SoC implementations continue to work and can be slowly ported over one by one. Like the ARM32 version it does not garbage collect page tables that become unused, so repeated mapping at different granularities will exhaust the available table space (this is presumed to be a reasonable limitation for a firmware environment and keeps the code much simpler). Also do some cleanup, align comments between coreboot and libpayload for easier diffing, and change all error cases to assert()s. Right now the code just propagates error codes up the stack until it eventually reaches a function that doesn't check them anymore. MMU configuration errors (essentially just misaligned requests and running out of table space) should always be compile-time programming errors, so failing hard and fast seems like the best way to deal with them. BRANCH=None BUG=None TEST=Compile-tested rush_ryu. Booted on Oak and hacked MMU init to use mmu_config_range() insted of memranges. Confirmed that CRCs over all page tables before and after the change are equal. Change-Id: I93585b44a277c1d96d31ee9c3dd2522b5e10085b Signed-off-by: Patrick Georgi <pgeorgi@chromium.org> Original-Commit-Id: f10fcba107aba1f3ea239471cb5a4f9239809539 Original-Change-Id: I6a2a11e3b94e6ae9e1553871f0cccd3b556b3e65 Original-Signed-off-by: Julius Werner <jwerner@chromium.org> Original-Reviewed-on: https://chromium-review.googlesource.com/271991 Original-Reviewed-by: Aaron Durbin <adurbin@chromium.org> Reviewed-on: http://review.coreboot.org/10304 Tested-by: build bot (Jenkins) Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net> Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
2015-05-18 22:11:12 +02:00
dsb();
isb();
}
arm64: Decouple MMU functions from memranges The current arm64 MMU interface is difficult to use in pre-RAM environments. It is based on the memranges API which makes use of malloc(), and early stages usually don't have a heap. It is also built as a one-shot interface that requires all memory ranges to be laid out beforehand, which is a problem when existing areas need to change (e.g. after initializing DRAM). The long-term goal of this patch is to completely switch to a configure-as-you-go interface based on the mmu_config_range() function, similar to what ARM32 does. As a first step this feature is added side-by-side to the existing interface so that existing SoC implementations continue to work and can be slowly ported over one by one. Like the ARM32 version it does not garbage collect page tables that become unused, so repeated mapping at different granularities will exhaust the available table space (this is presumed to be a reasonable limitation for a firmware environment and keeps the code much simpler). Also do some cleanup, align comments between coreboot and libpayload for easier diffing, and change all error cases to assert()s. Right now the code just propagates error codes up the stack until it eventually reaches a function that doesn't check them anymore. MMU configuration errors (essentially just misaligned requests and running out of table space) should always be compile-time programming errors, so failing hard and fast seems like the best way to deal with them. BRANCH=None BUG=None TEST=Compile-tested rush_ryu. Booted on Oak and hacked MMU init to use mmu_config_range() insted of memranges. Confirmed that CRCs over all page tables before and after the change are equal. Change-Id: I93585b44a277c1d96d31ee9c3dd2522b5e10085b Signed-off-by: Patrick Georgi <pgeorgi@chromium.org> Original-Commit-Id: f10fcba107aba1f3ea239471cb5a4f9239809539 Original-Change-Id: I6a2a11e3b94e6ae9e1553871f0cccd3b556b3e65 Original-Signed-off-by: Julius Werner <jwerner@chromium.org> Original-Reviewed-on: https://chromium-review.googlesource.com/271991 Original-Reviewed-by: Aaron Durbin <adurbin@chromium.org> Reviewed-on: http://review.coreboot.org/10304 Tested-by: build bot (Jenkins) Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net> Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
2015-05-18 22:11:12 +02:00
/* Func : mmu_init
* Desc : Initialize mmu based on the mmu_memrange passed. ttb_buffer is used as
* the base address for xlat tables. TTB_DEFAULT_SIZE defines the max number of
* tables that can be used
* Assuming that memory 0-4GiB is device memory.
*/
uint64_t mmu_init(struct mmu_ranges *mmu_ranges)
{
int i = 0;
xlat_addr = (uint64_t *)&ttb_buffer;
memset((void*)xlat_addr, 0, GRANULE_SIZE);
max_tables = (TTB_DEFAULT_SIZE >> GRANULE_SIZE_SHIFT);
free_idx = 1;
printf("Libpayload ARM64: TTB_BUFFER: %p Max Tables: %d\n",
(void*)xlat_addr, max_tables);
/*
* To keep things simple we start with mapping the entire base 4GB as
* device memory. This accommodates various architectures' default
* settings (for instance rk3399 mmio starts at 0xf8000000); it is
* fine tuned (e.g. mapping DRAM areas as write-back) later in the
* boot process.
*/
mmu_config_range(NULL, 0x100000000, TYPE_DEV_MEM);
arm64: Decouple MMU functions from memranges The current arm64 MMU interface is difficult to use in pre-RAM environments. It is based on the memranges API which makes use of malloc(), and early stages usually don't have a heap. It is also built as a one-shot interface that requires all memory ranges to be laid out beforehand, which is a problem when existing areas need to change (e.g. after initializing DRAM). The long-term goal of this patch is to completely switch to a configure-as-you-go interface based on the mmu_config_range() function, similar to what ARM32 does. As a first step this feature is added side-by-side to the existing interface so that existing SoC implementations continue to work and can be slowly ported over one by one. Like the ARM32 version it does not garbage collect page tables that become unused, so repeated mapping at different granularities will exhaust the available table space (this is presumed to be a reasonable limitation for a firmware environment and keeps the code much simpler). Also do some cleanup, align comments between coreboot and libpayload for easier diffing, and change all error cases to assert()s. Right now the code just propagates error codes up the stack until it eventually reaches a function that doesn't check them anymore. MMU configuration errors (essentially just misaligned requests and running out of table space) should always be compile-time programming errors, so failing hard and fast seems like the best way to deal with them. BRANCH=None BUG=None TEST=Compile-tested rush_ryu. Booted on Oak and hacked MMU init to use mmu_config_range() insted of memranges. Confirmed that CRCs over all page tables before and after the change are equal. Change-Id: I93585b44a277c1d96d31ee9c3dd2522b5e10085b Signed-off-by: Patrick Georgi <pgeorgi@chromium.org> Original-Commit-Id: f10fcba107aba1f3ea239471cb5a4f9239809539 Original-Change-Id: I6a2a11e3b94e6ae9e1553871f0cccd3b556b3e65 Original-Signed-off-by: Julius Werner <jwerner@chromium.org> Original-Reviewed-on: https://chromium-review.googlesource.com/271991 Original-Reviewed-by: Aaron Durbin <adurbin@chromium.org> Reviewed-on: http://review.coreboot.org/10304 Tested-by: build bot (Jenkins) Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net> Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
2015-05-18 22:11:12 +02:00
for (; i < mmu_ranges->used; i++)
mmu_config_range((void *)mmu_ranges->entries[i].base,
mmu_ranges->entries[i].size,
mmu_ranges->entries[i].type);
printf("Libpayload ARM64: MMU init done\n");
return 0;
}
static uint32_t is_mmu_enabled(void)
{
uint32_t sctlr;
sctlr = raw_read_sctlr_el2();
return (sctlr & SCTLR_M);
}
/*
* Func: mmu_disable
* Desc: Invalidate caches and disable mmu
*/
void mmu_disable(void)
{
uint32_t sctlr;
sctlr = raw_read_sctlr_el2();
sctlr &= ~(SCTLR_C | SCTLR_M | SCTLR_I);
tlbiall_el2();
dcache_clean_invalidate_all();
dsb();
isb();
raw_write_sctlr_el2(sctlr);
dcache_clean_invalidate_all();
dsb();
isb();
}
/*
* Func: mmu_enable
* Desc: Initialize MAIR, TCR, TTBR and enable MMU by setting appropriate bits
* in SCTLR
*/
void mmu_enable(void)
{
uint32_t sctlr;
/* Initialize MAIR indices */
raw_write_mair_el2(MAIR_ATTRIBUTES);
/* Invalidate TLBs */
tlbiall_el2();
/* Initialize TCR flags */
raw_write_tcr_el2(TCR_TOSZ | TCR_IRGN0_NM_WBWAC | TCR_ORGN0_NM_WBWAC |
TCR_SH0_IS | TCR_TG0_4KB | TCR_PS_256TB |
TCR_TBI_USED);
/* Initialize TTBR */
raw_write_ttbr0_el2((uintptr_t)xlat_addr);
arm64: Decouple MMU functions from memranges The current arm64 MMU interface is difficult to use in pre-RAM environments. It is based on the memranges API which makes use of malloc(), and early stages usually don't have a heap. It is also built as a one-shot interface that requires all memory ranges to be laid out beforehand, which is a problem when existing areas need to change (e.g. after initializing DRAM). The long-term goal of this patch is to completely switch to a configure-as-you-go interface based on the mmu_config_range() function, similar to what ARM32 does. As a first step this feature is added side-by-side to the existing interface so that existing SoC implementations continue to work and can be slowly ported over one by one. Like the ARM32 version it does not garbage collect page tables that become unused, so repeated mapping at different granularities will exhaust the available table space (this is presumed to be a reasonable limitation for a firmware environment and keeps the code much simpler). Also do some cleanup, align comments between coreboot and libpayload for easier diffing, and change all error cases to assert()s. Right now the code just propagates error codes up the stack until it eventually reaches a function that doesn't check them anymore. MMU configuration errors (essentially just misaligned requests and running out of table space) should always be compile-time programming errors, so failing hard and fast seems like the best way to deal with them. BRANCH=None BUG=None TEST=Compile-tested rush_ryu. Booted on Oak and hacked MMU init to use mmu_config_range() insted of memranges. Confirmed that CRCs over all page tables before and after the change are equal. Change-Id: I93585b44a277c1d96d31ee9c3dd2522b5e10085b Signed-off-by: Patrick Georgi <pgeorgi@chromium.org> Original-Commit-Id: f10fcba107aba1f3ea239471cb5a4f9239809539 Original-Change-Id: I6a2a11e3b94e6ae9e1553871f0cccd3b556b3e65 Original-Signed-off-by: Julius Werner <jwerner@chromium.org> Original-Reviewed-on: https://chromium-review.googlesource.com/271991 Original-Reviewed-by: Aaron Durbin <adurbin@chromium.org> Reviewed-on: http://review.coreboot.org/10304 Tested-by: build bot (Jenkins) Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net> Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
2015-05-18 22:11:12 +02:00
/* Ensure system register writes are committed before enabling MMU */
isb();
/* Enable MMU */
sctlr = raw_read_sctlr_el2();
sctlr |= SCTLR_C | SCTLR_M | SCTLR_I;
raw_write_sctlr_el2(sctlr);
isb();
if(is_mmu_enabled())
printf("ARM64: MMU enable done\n");
else
printf("ARM64: MMU enable failed\n");
}
/*
* Func: mmu_add_memrange
* Desc: Adds a new memory range
*/
static struct mmu_memrange *mmu_add_memrange(struct mmu_ranges *r,
uint64_t base, uint64_t size,
uint64_t type)
{
struct mmu_memrange *curr = NULL;
int i = r->used;
if (i < ARRAY_SIZE(r->entries)) {
curr = &r->entries[i];
curr->base = base;
curr->size = size;
curr->type = type;
r->used = i + 1;
}
return curr;
}
/* Structure to define properties of new memrange request */
struct mmu_new_range_prop {
/* Type of memrange */
uint64_t type;
/* Size of the range */
uint64_t size;
/*
* If any restrictions on the max addr limit(This addr is exclusive for
* the range), else 0
*/
uint64_t lim_excl;
/* If any restrictions on alignment of the range base, else 0 */
uint64_t align;
/*
* Function to test whether selected range is fine.
* NULL=any range is fine
* Return value 1=valid range, 0=otherwise
*/
int (*is_valid_range)(uint64_t, uint64_t);
/* From what type of source range should this range be extracted */
uint64_t src_type;
};
/*
* Func: mmu_is_range_free
* Desc: We need to ensure that the new range being allocated doesnt overlap
* with any used memory range. Basically:
* 1. Memory ranges used by the payload (usedmem_ranges)
* 2. Any area that falls below _end symbol in linker script (Kernel needs to be
* loaded in lower areas of memory, So, the payload linker script can have
* kernel memory below _start and _end. Thus, we want to make sure we do not
* step in those areas as well.
* Returns: 1 on success, 0 on error
* ASSUMPTION: All the memory used by payload resides below the program
* proper. If there is any memory used above the _end symbol, then it should be
* marked as used memory in usedmem_ranges during the presysinfo_scan.
*/
static int mmu_is_range_free(uint64_t r_base,
uint64_t r_end)
{
uint64_t payload_end = (uint64_t)&_end;
uint64_t i;
struct mmu_memrange *r = &usedmem_ranges.entries[0];
/* Allocate memranges only above payload */
if ((r_base <= payload_end) || (r_end <= payload_end))
return 0;
for (i = 0; i < usedmem_ranges.used; i++) {
uint64_t start = r[i].base;
uint64_t end = start + r[i].size;
if ((start < r_end) && (end > r_base))
return 0;
}
return 1;
}
/*
* Func: mmu_get_new_range
* Desc: Add a requested new memrange. We take as input set of all memranges and
* a structure to define the new memrange properties i.e. its type, size,
* max_addr it can grow upto, alignment restrictions, source type to take range
* from and finally a function pointer to check if the chosen range is valid.
*/
static struct mmu_memrange *mmu_get_new_range(struct mmu_ranges *mmu_ranges,
struct mmu_new_range_prop *new)
{
int i = 0;
struct mmu_memrange *r = &mmu_ranges->entries[0];
if (new->size == 0) {
printf("MMU Error: Invalid range size\n");
return NULL;
}
for (; i < mmu_ranges->used; i++) {
if ((r[i].type != new->src_type) ||
(r[i].size < new->size) ||
(new->lim_excl && (r[i].base >= new->lim_excl)))
continue;
uint64_t base_addr;
uint64_t range_end_addr = r[i].base + r[i].size;
uint64_t end_addr = range_end_addr;
/* Make sure we do not go above max if it is non-zero */
if (new->lim_excl && (end_addr >= new->lim_excl))
end_addr = new->lim_excl;
while (1) {
/*
* In case of alignment requirement,
* if end_addr is aligned, then base_addr will be too.
*/
if (new->align)
end_addr = ALIGN_DOWN(end_addr, new->align);
base_addr = end_addr - new->size;
if (base_addr < r[i].base)
break;
/*
* If the selected range is not used and valid for the
* user, move ahead with it
*/
if (mmu_is_range_free(base_addr, end_addr) &&
((new->is_valid_range == NULL) ||
new->is_valid_range(base_addr, end_addr)))
break;
/* Drop to the next address. */
end_addr -= 1;
}
if (base_addr < r[i].base)
continue;
if (end_addr != range_end_addr) {
/* Add a new memrange since we split up one
* range crossing the 4GiB boundary or doing an
* ALIGN_DOWN on end_addr.
*/
r[i].size -= (range_end_addr - end_addr);
if (mmu_add_memrange(mmu_ranges, end_addr,
range_end_addr - end_addr,
r[i].type) == NULL)
mmu_error();
}
if (r[i].size == new->size) {
r[i].type = new->type;
return &r[i];
}
r[i].size -= new->size;
r = mmu_add_memrange(mmu_ranges, base_addr, new->size,
new->type);
if (r == NULL)
mmu_error();
return r;
}
/* Should never reach here if everything went fine */
printf("ARM64 ERROR: No region allocated\n");
return NULL;
}
/*
* Func: mmu_alloc_range
* Desc: Call get_new_range to get a new memrange which is unused and mark it as
* used to avoid same range being allocated for different purposes.
*/
static struct mmu_memrange *mmu_alloc_range(struct mmu_ranges *mmu_ranges,
struct mmu_new_range_prop *p)
{
struct mmu_memrange *r = mmu_get_new_range(mmu_ranges, p);
if (r == NULL)
return NULL;
/*
* Mark this memrange as used memory. Important since function
* can be called multiple times and we do not want to reuse some
* range already allocated.
*/
if (mmu_add_memrange(&usedmem_ranges, r->base, r->size, r->type)
== NULL)
mmu_error();
return r;
}
/*
* Func: mmu_add_dma_range
* Desc: Add a memrange for dma operations. This is special because we want to
* initialize this memory as non-cacheable. We have a constraint that the DMA
* buffer should be below 4GiB(32-bit only). So, we lookup a TYPE_NORMAL_MEM
* from the lowest available addresses and align it to page size i.e. 64KiB.
*/
static struct mmu_memrange *mmu_add_dma_range(struct mmu_ranges *mmu_ranges)
{
struct mmu_new_range_prop prop;
prop.type = TYPE_DMA_MEM;
/* DMA_DEFAULT_SIZE is multiple of GRANULE_SIZE */
assert((DMA_DEFAULT_SIZE % GRANULE_SIZE) == 0);
prop.size = DMA_DEFAULT_SIZE;
prop.lim_excl = (uint64_t)CONFIG_LP_DMA_LIM_EXCL * MiB;
prop.align = GRANULE_SIZE;
prop.is_valid_range = NULL;
prop.src_type = TYPE_NORMAL_MEM;
return mmu_alloc_range(mmu_ranges, &prop);
}
static struct mmu_memrange *_mmu_add_fb_range(
uint32_t size,
struct mmu_ranges *mmu_ranges)
{
struct mmu_new_range_prop prop;
prop.type = TYPE_DMA_MEM;
/* make sure to allocate a size of multiple of GRANULE_SIZE */
size = ALIGN_UP(size, GRANULE_SIZE);
prop.size = size;
prop.lim_excl = MIN_64_BIT_ADDR;
prop.align = MB_SIZE;
prop.is_valid_range = NULL;
prop.src_type = TYPE_NORMAL_MEM;
return mmu_alloc_range(mmu_ranges, &prop);
}
/*
* Func: mmu_extract_ranges
* Desc: Assumption is that coreboot tables have memranges in sorted
* order. So, if there is an opportunity to combine ranges, we do that as
* well. Memranges are initialized for both CB_MEM_RAM and CB_MEM_TABLE as
* TYPE_NORMAL_MEM.
*/
static void mmu_extract_ranges(struct memrange *cb_ranges,
uint64_t ncb,
struct mmu_ranges *mmu_ranges)
{
int i = 0;
struct mmu_memrange *prev_range = NULL;
/* Extract memory ranges to be mapped */
for (; i < ncb; i++) {
switch (cb_ranges[i].type) {
case CB_MEM_RAM:
case CB_MEM_TABLE:
if (prev_range && (prev_range->base + prev_range->size
== cb_ranges[i].base)) {
prev_range->size += cb_ranges[i].size;
} else {
prev_range = mmu_add_memrange(mmu_ranges,
cb_ranges[i].base,
cb_ranges[i].size,
TYPE_NORMAL_MEM);
if (prev_range == NULL)
mmu_error();
}
break;
default:
break;
}
}
}
static void mmu_add_fb_range(struct mmu_ranges *mmu_ranges)
{
struct mmu_memrange *fb_range;
static struct cb_framebuffer modified_fb;
struct cb_framebuffer *framebuffer = lib_sysinfo.framebuffer;
uint32_t fb_size;
/* Check whether framebuffer is needed */
if (framebuffer == NULL)
return;
fb_size = framebuffer->bytes_per_line * framebuffer->y_resolution;
if (!fb_size)
return;
/* framebuffer address has been set already, so just add it as DMA */
if (framebuffer->physical_address) {
if (mmu_add_memrange(mmu_ranges,
framebuffer->physical_address,
fb_size,
TYPE_DMA_MEM) == NULL)
mmu_error();
return;
}
/* Allocate framebuffer */
fb_range = _mmu_add_fb_range(fb_size, mmu_ranges);
if (fb_range == NULL)
mmu_error();
/*
* Set framebuffer address. However, one needs to use a freshly
* allocated framebuffer structure because the one in the coreboot
* table is part of a checksum calculation. Therefore, one cannot
* modify a field without recomputing the necessary checksum
* calcuation.
*/
modified_fb = *framebuffer;
modified_fb.physical_address = fb_range->base;
lib_sysinfo.framebuffer = &modified_fb;
}
/*
* Func: mmu_init_ranges
* Desc: Initialize mmu_memranges based on the memranges obtained from coreboot
* tables. Also, initialize dma memrange and xlat_addr for ttb buffer.
*/
struct mmu_memrange *mmu_init_ranges_from_sysinfo(struct memrange *cb_ranges,
uint64_t ncb,
struct mmu_ranges *mmu_ranges)
{
struct mmu_memrange *dma_range;
/* Initialize mmu_ranges to contain no entries. */
mmu_ranges->used = 0;
/* Extract ranges from memrange in lib_sysinfo */
mmu_extract_ranges(cb_ranges, ncb, mmu_ranges);
/* Get a range for dma */
dma_range = mmu_add_dma_range(mmu_ranges);
/* Get a range for framebuffer */
mmu_add_fb_range(mmu_ranges);
if (dma_range == NULL)
mmu_error();
return dma_range;
}
/*
* Func: mmu_presysinfo_memory_used
* Desc: Initializes all the memory used for presysinfo page table
* initialization and enabling of MMU. All these ranges are stored in
* usedmem_ranges. usedmem_ranges plays an important role in selecting the dma
* buffer as well since we check the dma buffer range against the used memory
* ranges to prevent any overstepping.
*/
void mmu_presysinfo_memory_used(uint64_t base, uint64_t size)
{
uint64_t range_base;
range_base = ALIGN_DOWN(base, GRANULE_SIZE);
size += (base - range_base);
size = ALIGN_UP(size, GRANULE_SIZE);
mmu_add_memrange(&usedmem_ranges, range_base, size, TYPE_NORMAL_MEM);
}
void mmu_presysinfo_enable(void)
{
mmu_init(&usedmem_ranges);
mmu_enable();
}