linux/arch/riscv/kernel/setup.c
Palmer Dabbelt 52b77c2806
Merge patch series "riscv: Reduce ARCH_KMALLOC_MINALIGN to 8"
Jisheng Zhang <jszhang@kernel.org> says:

Currently, riscv defines ARCH_DMA_MINALIGN as L1_CACHE_BYTES, I.E
64Bytes, if CONFIG_RISCV_DMA_NONCOHERENT=y. To support unified kernel
Image, usually we have to enable CONFIG_RISCV_DMA_NONCOHERENT, thus
it brings some bad effects to coherent platforms:

Firstly, it wastes memory, kmalloc-96, kmalloc-32, kmalloc-16 and
kmalloc-8 slab caches don't exist any more, they are replaced with
either kmalloc-128 or kmalloc-64.

Secondly, larger than necessary kmalloc aligned allocations results
in unnecessary cache/TLB pressure.

This issue also exists on arm64 platforms. From last year, Catalin
tried to solve this issue by decoupling ARCH_KMALLOC_MINALIGN from
ARCH_DMA_MINALIGN, limiting kmalloc() minimum alignment to
dma_get_cache_alignment() and replacing ARCH_KMALLOC_MINALIGN usage
in various drivers with ARCH_DMA_MINALIGN etc.[1]

One fact we can make use of for riscv: if the CPU doesn't support
ZICBOM or T-HEAD CMO, we know the platform is coherent. Based on
Catalin's work and above fact, we can easily solve the kmalloc align
issue for riscv: we can override dma_get_cache_alignment(), then let
it return ARCH_DMA_MINALIGN at the beginning and return 1 once we know
the underlying HW neither supports ZICBOM nor supports T-HEAD CMO.

So what about if the CPU supports ZICBOM or T-HEAD CMO, but all the
devices are dma coherent? Well, we use ARCH_DMA_MINALIGN as the
kmalloc minimum alignment, nothing changed in this case. This case
can be improved in the future once we see such platforms in mainline.

After this patch, a simple test of booting to a small buildroot rootfs
on qemu shows:

kmalloc-96           5041    5041     96  ...
kmalloc-64           9606    9606     64  ...
kmalloc-32           5128    5128     32  ...
kmalloc-16           7682    7682     16  ...
kmalloc-8           10246   10246      8  ...

So we save about 1268KB memory. The saving will be much larger in normal
OS env on real HW platforms.

patch1 allows kmalloc() caches aligned to the smallest value.
patch2 enables DMA_BOUNCE_UNALIGNED_KMALLOC.

After this series:

As for coherent platforms, kmalloc-{8,16,32,96} caches come back on
coherent both RV32 and RV64 platforms, I.E !ZICBOM and !THEAD_CMO.

As for noncoherent RV32 platforms, nothing changed.

As for noncoherent RV64 platforms, I.E either ZICBOM or THEAD_CMO, the
above kmalloc caches also come back if > 4GB memory or users pass
"swiotlb=mmnn,force" to force swiotlb creation if <= 4GB memory. How
much mmnn should be depends on the specific platform, it needs to be
tried and tested all possible usage case on the specific hardware. For
example, I can use the minimal I/O TLB slabs on Sipeed M1S Dock.

* b4-shazam-merge:
  riscv: enable DMA_BOUNCE_UNALIGNED_KMALLOC for !dma_coherent
  riscv: allow kmalloc() caches aligned to the smallest value

Link: https://lore.kernel.org/linux-arm-kernel/20230524171904.3967031-1-catalin.marinas@arm.com/ [1]
Link: https://lore.kernel.org/r/20230718152214.2907-1-jszhang@kernel.org
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
2023-08-31 00:18:35 -07:00

350 lines
8.6 KiB
C

// SPDX-License-Identifier: GPL-2.0-or-later
/*
* Copyright (C) 2009 Sunplus Core Technology Co., Ltd.
* Chen Liqin <liqin.chen@sunplusct.com>
* Lennox Wu <lennox.wu@sunplusct.com>
* Copyright (C) 2012 Regents of the University of California
* Copyright (C) 2020 FORTH-ICS/CARV
* Nick Kossifidis <mick@ics.forth.gr>
*/
#include <linux/acpi.h>
#include <linux/cpu.h>
#include <linux/init.h>
#include <linux/mm.h>
#include <linux/memblock.h>
#include <linux/sched.h>
#include <linux/console.h>
#include <linux/screen_info.h>
#include <linux/of_fdt.h>
#include <linux/sched/task.h>
#include <linux/smp.h>
#include <linux/efi.h>
#include <linux/crash_dump.h>
#include <asm/acpi.h>
#include <asm/alternative.h>
#include <asm/cacheflush.h>
#include <asm/cpu_ops.h>
#include <asm/early_ioremap.h>
#include <asm/pgtable.h>
#include <asm/setup.h>
#include <asm/set_memory.h>
#include <asm/sections.h>
#include <asm/sbi.h>
#include <asm/tlbflush.h>
#include <asm/thread_info.h>
#include <asm/kasan.h>
#include <asm/efi.h>
#include "head.h"
#if defined(CONFIG_DUMMY_CONSOLE) || defined(CONFIG_EFI)
struct screen_info screen_info __section(".data") = {
.orig_video_lines = 30,
.orig_video_cols = 80,
.orig_video_mode = 0,
.orig_video_ega_bx = 0,
.orig_video_isVGA = 1,
.orig_video_points = 8
};
#endif
/*
* The lucky hart to first increment this variable will boot the other cores.
* This is used before the kernel initializes the BSS so it can't be in the
* BSS.
*/
atomic_t hart_lottery __section(".sdata")
#ifdef CONFIG_XIP_KERNEL
= ATOMIC_INIT(0xC001BEEF)
#endif
;
unsigned long boot_cpu_hartid;
static DEFINE_PER_CPU(struct cpu, cpu_devices);
/*
* Place kernel memory regions on the resource tree so that
* kexec-tools can retrieve them from /proc/iomem. While there
* also add "System RAM" regions for compatibility with other
* archs, and the rest of the known regions for completeness.
*/
static struct resource kimage_res = { .name = "Kernel image", };
static struct resource code_res = { .name = "Kernel code", };
static struct resource data_res = { .name = "Kernel data", };
static struct resource rodata_res = { .name = "Kernel rodata", };
static struct resource bss_res = { .name = "Kernel bss", };
#ifdef CONFIG_CRASH_DUMP
static struct resource elfcorehdr_res = { .name = "ELF Core hdr", };
#endif
static int __init add_resource(struct resource *parent,
struct resource *res)
{
int ret = 0;
ret = insert_resource(parent, res);
if (ret < 0) {
pr_err("Failed to add a %s resource at %llx\n",
res->name, (unsigned long long) res->start);
return ret;
}
return 1;
}
static int __init add_kernel_resources(void)
{
int ret = 0;
/*
* The memory region of the kernel image is continuous and
* was reserved on setup_bootmem, register it here as a
* resource, with the various segments of the image as
* child nodes.
*/
code_res.start = __pa_symbol(_text);
code_res.end = __pa_symbol(_etext) - 1;
code_res.flags = IORESOURCE_SYSTEM_RAM | IORESOURCE_BUSY;
rodata_res.start = __pa_symbol(__start_rodata);
rodata_res.end = __pa_symbol(__end_rodata) - 1;
rodata_res.flags = IORESOURCE_SYSTEM_RAM | IORESOURCE_BUSY;
data_res.start = __pa_symbol(_data);
data_res.end = __pa_symbol(_edata) - 1;
data_res.flags = IORESOURCE_SYSTEM_RAM | IORESOURCE_BUSY;
bss_res.start = __pa_symbol(__bss_start);
bss_res.end = __pa_symbol(__bss_stop) - 1;
bss_res.flags = IORESOURCE_SYSTEM_RAM | IORESOURCE_BUSY;
kimage_res.start = code_res.start;
kimage_res.end = bss_res.end;
kimage_res.flags = IORESOURCE_SYSTEM_RAM | IORESOURCE_BUSY;
ret = add_resource(&iomem_resource, &kimage_res);
if (ret < 0)
return ret;
ret = add_resource(&kimage_res, &code_res);
if (ret < 0)
return ret;
ret = add_resource(&kimage_res, &rodata_res);
if (ret < 0)
return ret;
ret = add_resource(&kimage_res, &data_res);
if (ret < 0)
return ret;
ret = add_resource(&kimage_res, &bss_res);
return ret;
}
static void __init init_resources(void)
{
struct memblock_region *region = NULL;
struct resource *res = NULL;
struct resource *mem_res = NULL;
size_t mem_res_sz = 0;
int num_resources = 0, res_idx = 0;
int ret = 0;
/* + 1 as memblock_alloc() might increase memblock.reserved.cnt */
num_resources = memblock.memory.cnt + memblock.reserved.cnt + 1;
res_idx = num_resources - 1;
mem_res_sz = num_resources * sizeof(*mem_res);
mem_res = memblock_alloc(mem_res_sz, SMP_CACHE_BYTES);
if (!mem_res)
panic("%s: Failed to allocate %zu bytes\n", __func__, mem_res_sz);
/*
* Start by adding the reserved regions, if they overlap
* with /memory regions, insert_resource later on will take
* care of it.
*/
ret = add_kernel_resources();
if (ret < 0)
goto error;
#ifdef CONFIG_KEXEC_CORE
if (crashk_res.start != crashk_res.end) {
ret = add_resource(&iomem_resource, &crashk_res);
if (ret < 0)
goto error;
}
if (crashk_low_res.start != crashk_low_res.end) {
ret = add_resource(&iomem_resource, &crashk_low_res);
if (ret < 0)
goto error;
}
#endif
#ifdef CONFIG_CRASH_DUMP
if (elfcorehdr_size > 0) {
elfcorehdr_res.start = elfcorehdr_addr;
elfcorehdr_res.end = elfcorehdr_addr + elfcorehdr_size - 1;
elfcorehdr_res.flags = IORESOURCE_SYSTEM_RAM | IORESOURCE_BUSY;
add_resource(&iomem_resource, &elfcorehdr_res);
}
#endif
for_each_reserved_mem_region(region) {
res = &mem_res[res_idx--];
res->name = "Reserved";
res->flags = IORESOURCE_MEM | IORESOURCE_EXCLUSIVE;
res->start = __pfn_to_phys(memblock_region_reserved_base_pfn(region));
res->end = __pfn_to_phys(memblock_region_reserved_end_pfn(region)) - 1;
/*
* Ignore any other reserved regions within
* system memory.
*/
if (memblock_is_memory(res->start)) {
/* Re-use this pre-allocated resource */
res_idx++;
continue;
}
ret = add_resource(&iomem_resource, res);
if (ret < 0)
goto error;
}
/* Add /memory regions to the resource tree */
for_each_mem_region(region) {
res = &mem_res[res_idx--];
if (unlikely(memblock_is_nomap(region))) {
res->name = "Reserved";
res->flags = IORESOURCE_MEM | IORESOURCE_EXCLUSIVE;
} else {
res->name = "System RAM";
res->flags = IORESOURCE_SYSTEM_RAM | IORESOURCE_BUSY;
}
res->start = __pfn_to_phys(memblock_region_memory_base_pfn(region));
res->end = __pfn_to_phys(memblock_region_memory_end_pfn(region)) - 1;
ret = add_resource(&iomem_resource, res);
if (ret < 0)
goto error;
}
/* Clean-up any unused pre-allocated resources */
if (res_idx >= 0)
memblock_free(mem_res, (res_idx + 1) * sizeof(*mem_res));
return;
error:
/* Better an empty resource tree than an inconsistent one */
release_child_resources(&iomem_resource);
memblock_free(mem_res, mem_res_sz);
}
static void __init parse_dtb(void)
{
/* Early scan of device tree from init memory */
if (early_init_dt_scan(dtb_early_va)) {
const char *name = of_flat_dt_get_machine_name();
if (name) {
pr_info("Machine model: %s\n", name);
dump_stack_set_arch_desc("%s (DT)", name);
}
} else {
pr_err("No DTB passed to the kernel\n");
}
#ifdef CONFIG_CMDLINE_FORCE
strscpy(boot_command_line, CONFIG_CMDLINE, COMMAND_LINE_SIZE);
pr_info("Forcing kernel command line to: %s\n", boot_command_line);
#endif
}
extern void __init init_rt_signal_env(void);
void __init setup_arch(char **cmdline_p)
{
parse_dtb();
setup_initial_init_mm(_stext, _etext, _edata, _end);
*cmdline_p = boot_command_line;
early_ioremap_setup();
sbi_init();
jump_label_init();
parse_early_param();
efi_init();
paging_init();
/* Parse the ACPI tables for possible boot-time configuration */
acpi_boot_table_init();
#if IS_ENABLED(CONFIG_BUILTIN_DTB)
unflatten_and_copy_device_tree();
#else
unflatten_device_tree();
#endif
misc_mem_init();
init_resources();
#ifdef CONFIG_KASAN
kasan_init();
#endif
#ifdef CONFIG_SMP
setup_smp();
#endif
if (!acpi_disabled)
acpi_init_rintc_map();
riscv_init_cbo_blocksizes();
riscv_fill_hwcap();
init_rt_signal_env();
apply_boot_alternatives();
if (IS_ENABLED(CONFIG_RISCV_ISA_ZICBOM) &&
riscv_isa_extension_available(NULL, ZICBOM))
riscv_noncoherent_supported();
riscv_set_dma_cache_alignment();
}
static int __init topology_init(void)
{
int i, ret;
for_each_possible_cpu(i) {
struct cpu *cpu = &per_cpu(cpu_devices, i);
cpu->hotpluggable = cpu_has_hotplug(i);
ret = register_cpu(cpu, i);
if (unlikely(ret))
pr_warn("Warning: %s: register_cpu %d failed (%d)\n",
__func__, i, ret);
}
return 0;
}
subsys_initcall(topology_init);
void free_initmem(void)
{
if (IS_ENABLED(CONFIG_STRICT_KERNEL_RWX)) {
set_kernel_memory(lm_alias(__init_begin), lm_alias(__init_end), set_memory_rw_nx);
if (IS_ENABLED(CONFIG_64BIT))
set_kernel_memory(__init_begin, __init_end, set_memory_nx);
}
free_initmem_default(POISON_FREE_INITMEM);
}