bcf3d957c6
Conceptually, we want the memory mappings to always be up to date and represent whatever is in the TLB. To ensure that, we need to sync them over in the userspace case and for the kernel we need to process the mappings. The kernel will call flush_tlb_* if page table entries that were valid before become invalid. Unfortunately, this is not the case if entries are added. As such, change both flush_tlb_* and set_ptes to track the memory range that has to be synchronized. For the kernel, we need to execute a flush_tlb_kern_* immediately but we can wait for the first page fault in case of set_ptes. For userspace in contrast we only store that a range of memory needs to be synced and do so whenever we switch to that process. Signed-off-by: Benjamin Berg <benjamin.berg@intel.com> Link: https://patch.msgid.link/20240703134536.1161108-13-benjamin@sipsolutions.net Signed-off-by: Johannes Berg <johannes.berg@intel.com>
325 lines
8.2 KiB
C
325 lines
8.2 KiB
C
// SPDX-License-Identifier: GPL-2.0
|
|
/*
|
|
* Copyright (C) 2000 - 2007 Jeff Dike (jdike@{addtoit,linux.intel}.com)
|
|
*/
|
|
|
|
#include <linux/mm.h>
|
|
#include <linux/sched/signal.h>
|
|
#include <linux/hardirq.h>
|
|
#include <linux/module.h>
|
|
#include <linux/uaccess.h>
|
|
#include <linux/sched/debug.h>
|
|
#include <asm/current.h>
|
|
#include <asm/tlbflush.h>
|
|
#include <arch.h>
|
|
#include <as-layout.h>
|
|
#include <kern_util.h>
|
|
#include <os.h>
|
|
#include <skas.h>
|
|
|
|
/*
|
|
* Note this is constrained to return 0, -EFAULT, -EACCES, -ENOMEM by
|
|
* segv().
|
|
*/
|
|
int handle_page_fault(unsigned long address, unsigned long ip,
|
|
int is_write, int is_user, int *code_out)
|
|
{
|
|
struct mm_struct *mm = current->mm;
|
|
struct vm_area_struct *vma;
|
|
pmd_t *pmd;
|
|
pte_t *pte;
|
|
int err = -EFAULT;
|
|
unsigned int flags = FAULT_FLAG_DEFAULT;
|
|
|
|
*code_out = SEGV_MAPERR;
|
|
|
|
/*
|
|
* If the fault was with pagefaults disabled, don't take the fault, just
|
|
* fail.
|
|
*/
|
|
if (faulthandler_disabled())
|
|
goto out_nosemaphore;
|
|
|
|
if (is_user)
|
|
flags |= FAULT_FLAG_USER;
|
|
retry:
|
|
mmap_read_lock(mm);
|
|
vma = find_vma(mm, address);
|
|
if (!vma)
|
|
goto out;
|
|
if (vma->vm_start <= address)
|
|
goto good_area;
|
|
if (!(vma->vm_flags & VM_GROWSDOWN))
|
|
goto out;
|
|
if (is_user && !ARCH_IS_STACKGROW(address))
|
|
goto out;
|
|
vma = expand_stack(mm, address);
|
|
if (!vma)
|
|
goto out_nosemaphore;
|
|
|
|
good_area:
|
|
*code_out = SEGV_ACCERR;
|
|
if (is_write) {
|
|
if (!(vma->vm_flags & VM_WRITE))
|
|
goto out;
|
|
flags |= FAULT_FLAG_WRITE;
|
|
} else {
|
|
/* Don't require VM_READ|VM_EXEC for write faults! */
|
|
if (!(vma->vm_flags & (VM_READ | VM_EXEC)))
|
|
goto out;
|
|
}
|
|
|
|
do {
|
|
vm_fault_t fault;
|
|
|
|
fault = handle_mm_fault(vma, address, flags, NULL);
|
|
|
|
if ((fault & VM_FAULT_RETRY) && fatal_signal_pending(current))
|
|
goto out_nosemaphore;
|
|
|
|
/* The fault is fully completed (including releasing mmap lock) */
|
|
if (fault & VM_FAULT_COMPLETED)
|
|
return 0;
|
|
|
|
if (unlikely(fault & VM_FAULT_ERROR)) {
|
|
if (fault & VM_FAULT_OOM) {
|
|
goto out_of_memory;
|
|
} else if (fault & VM_FAULT_SIGSEGV) {
|
|
goto out;
|
|
} else if (fault & VM_FAULT_SIGBUS) {
|
|
err = -EACCES;
|
|
goto out;
|
|
}
|
|
BUG();
|
|
}
|
|
if (fault & VM_FAULT_RETRY) {
|
|
flags |= FAULT_FLAG_TRIED;
|
|
|
|
goto retry;
|
|
}
|
|
|
|
pmd = pmd_off(mm, address);
|
|
pte = pte_offset_kernel(pmd, address);
|
|
} while (!pte_present(*pte));
|
|
err = 0;
|
|
/*
|
|
* The below warning was added in place of
|
|
* pte_mkyoung(); if (is_write) pte_mkdirty();
|
|
* If it's triggered, we'd see normally a hang here (a clean pte is
|
|
* marked read-only to emulate the dirty bit).
|
|
* However, the generic code can mark a PTE writable but clean on a
|
|
* concurrent read fault, triggering this harmlessly. So comment it out.
|
|
*/
|
|
#if 0
|
|
WARN_ON(!pte_young(*pte) || (is_write && !pte_dirty(*pte)));
|
|
#endif
|
|
|
|
out:
|
|
mmap_read_unlock(mm);
|
|
out_nosemaphore:
|
|
return err;
|
|
|
|
out_of_memory:
|
|
/*
|
|
* We ran out of memory, call the OOM killer, and return the userspace
|
|
* (which will retry the fault, or kill us if we got oom-killed).
|
|
*/
|
|
mmap_read_unlock(mm);
|
|
if (!is_user)
|
|
goto out_nosemaphore;
|
|
pagefault_out_of_memory();
|
|
return 0;
|
|
}
|
|
|
|
static void show_segv_info(struct uml_pt_regs *regs)
|
|
{
|
|
struct task_struct *tsk = current;
|
|
struct faultinfo *fi = UPT_FAULTINFO(regs);
|
|
|
|
if (!unhandled_signal(tsk, SIGSEGV))
|
|
return;
|
|
|
|
if (!printk_ratelimit())
|
|
return;
|
|
|
|
printk("%s%s[%d]: segfault at %lx ip %px sp %px error %x",
|
|
task_pid_nr(tsk) > 1 ? KERN_INFO : KERN_EMERG,
|
|
tsk->comm, task_pid_nr(tsk), FAULT_ADDRESS(*fi),
|
|
(void *)UPT_IP(regs), (void *)UPT_SP(regs),
|
|
fi->error_code);
|
|
|
|
print_vma_addr(KERN_CONT " in ", UPT_IP(regs));
|
|
printk(KERN_CONT "\n");
|
|
}
|
|
|
|
static void bad_segv(struct faultinfo fi, unsigned long ip)
|
|
{
|
|
current->thread.arch.faultinfo = fi;
|
|
force_sig_fault(SIGSEGV, SEGV_ACCERR, (void __user *) FAULT_ADDRESS(fi));
|
|
}
|
|
|
|
void fatal_sigsegv(void)
|
|
{
|
|
force_fatal_sig(SIGSEGV);
|
|
do_signal(¤t->thread.regs);
|
|
/*
|
|
* This is to tell gcc that we're not returning - do_signal
|
|
* can, in general, return, but in this case, it's not, since
|
|
* we just got a fatal SIGSEGV queued.
|
|
*/
|
|
os_dump_core();
|
|
}
|
|
|
|
/**
|
|
* segv_handler() - the SIGSEGV handler
|
|
* @sig: the signal number
|
|
* @unused_si: the signal info struct; unused in this handler
|
|
* @regs: the ptrace register information
|
|
*
|
|
* The handler first extracts the faultinfo from the UML ptrace regs struct.
|
|
* If the userfault did not happen in an UML userspace process, bad_segv is called.
|
|
* Otherwise the signal did happen in a cloned userspace process, handle it.
|
|
*/
|
|
void segv_handler(int sig, struct siginfo *unused_si, struct uml_pt_regs *regs)
|
|
{
|
|
struct faultinfo * fi = UPT_FAULTINFO(regs);
|
|
|
|
if (UPT_IS_USER(regs) && !SEGV_IS_FIXABLE(fi)) {
|
|
show_segv_info(regs);
|
|
bad_segv(*fi, UPT_IP(regs));
|
|
return;
|
|
}
|
|
segv(*fi, UPT_IP(regs), UPT_IS_USER(regs), regs);
|
|
}
|
|
|
|
/*
|
|
* We give a *copy* of the faultinfo in the regs to segv.
|
|
* This must be done, since nesting SEGVs could overwrite
|
|
* the info in the regs. A pointer to the info then would
|
|
* give us bad data!
|
|
*/
|
|
unsigned long segv(struct faultinfo fi, unsigned long ip, int is_user,
|
|
struct uml_pt_regs *regs)
|
|
{
|
|
jmp_buf *catcher;
|
|
int si_code;
|
|
int err;
|
|
int is_write = FAULT_WRITE(fi);
|
|
unsigned long address = FAULT_ADDRESS(fi);
|
|
|
|
if (!is_user && regs)
|
|
current->thread.segv_regs = container_of(regs, struct pt_regs, regs);
|
|
|
|
if (!is_user && init_mm.context.sync_tlb_range_to) {
|
|
/*
|
|
* Kernel has pending updates from set_ptes that were not
|
|
* flushed yet. Syncing them should fix the pagefault (if not
|
|
* we'll get here again and panic).
|
|
*/
|
|
err = um_tlb_sync(&init_mm);
|
|
if (err == -ENOMEM)
|
|
report_enomem();
|
|
if (err)
|
|
panic("Failed to sync kernel TLBs: %d", err);
|
|
goto out;
|
|
}
|
|
else if (current->mm == NULL) {
|
|
show_regs(container_of(regs, struct pt_regs, regs));
|
|
panic("Segfault with no mm");
|
|
}
|
|
else if (!is_user && address > PAGE_SIZE && address < TASK_SIZE) {
|
|
show_regs(container_of(regs, struct pt_regs, regs));
|
|
panic("Kernel tried to access user memory at addr 0x%lx, ip 0x%lx",
|
|
address, ip);
|
|
}
|
|
|
|
if (SEGV_IS_FIXABLE(&fi))
|
|
err = handle_page_fault(address, ip, is_write, is_user,
|
|
&si_code);
|
|
else {
|
|
err = -EFAULT;
|
|
/*
|
|
* A thread accessed NULL, we get a fault, but CR2 is invalid.
|
|
* This code is used in __do_copy_from_user() of TT mode.
|
|
* XXX tt mode is gone, so maybe this isn't needed any more
|
|
*/
|
|
address = 0;
|
|
}
|
|
|
|
catcher = current->thread.fault_catcher;
|
|
if (!err)
|
|
goto out;
|
|
else if (catcher != NULL) {
|
|
current->thread.fault_addr = (void *) address;
|
|
UML_LONGJMP(catcher, 1);
|
|
}
|
|
else if (current->thread.fault_addr != NULL)
|
|
panic("fault_addr set but no fault catcher");
|
|
else if (!is_user && arch_fixup(ip, regs))
|
|
goto out;
|
|
|
|
if (!is_user) {
|
|
show_regs(container_of(regs, struct pt_regs, regs));
|
|
panic("Kernel mode fault at addr 0x%lx, ip 0x%lx",
|
|
address, ip);
|
|
}
|
|
|
|
show_segv_info(regs);
|
|
|
|
if (err == -EACCES) {
|
|
current->thread.arch.faultinfo = fi;
|
|
force_sig_fault(SIGBUS, BUS_ADRERR, (void __user *)address);
|
|
} else {
|
|
BUG_ON(err != -EFAULT);
|
|
current->thread.arch.faultinfo = fi;
|
|
force_sig_fault(SIGSEGV, si_code, (void __user *) address);
|
|
}
|
|
|
|
out:
|
|
if (regs)
|
|
current->thread.segv_regs = NULL;
|
|
|
|
return 0;
|
|
}
|
|
|
|
void relay_signal(int sig, struct siginfo *si, struct uml_pt_regs *regs)
|
|
{
|
|
int code, err;
|
|
if (!UPT_IS_USER(regs)) {
|
|
if (sig == SIGBUS)
|
|
printk(KERN_ERR "Bus error - the host /dev/shm or /tmp "
|
|
"mount likely just ran out of space\n");
|
|
panic("Kernel mode signal %d", sig);
|
|
}
|
|
|
|
arch_examine_signal(sig, regs);
|
|
|
|
/* Is the signal layout for the signal known?
|
|
* Signal data must be scrubbed to prevent information leaks.
|
|
*/
|
|
code = si->si_code;
|
|
err = si->si_errno;
|
|
if ((err == 0) && (siginfo_layout(sig, code) == SIL_FAULT)) {
|
|
struct faultinfo *fi = UPT_FAULTINFO(regs);
|
|
current->thread.arch.faultinfo = *fi;
|
|
force_sig_fault(sig, code, (void __user *)FAULT_ADDRESS(*fi));
|
|
} else {
|
|
printk(KERN_ERR "Attempted to relay unknown signal %d (si_code = %d) with errno %d\n",
|
|
sig, code, err);
|
|
force_sig(sig);
|
|
}
|
|
}
|
|
|
|
void bus_handler(int sig, struct siginfo *si, struct uml_pt_regs *regs)
|
|
{
|
|
if (current->thread.fault_catcher != NULL)
|
|
UML_LONGJMP(current->thread.fault_catcher, 1);
|
|
else
|
|
relay_signal(sig, si, regs);
|
|
}
|
|
|
|
void winch(int sig, struct siginfo *unused_si, struct uml_pt_regs *regs)
|
|
{
|
|
do_IRQ(WINCH_IRQ, regs);
|
|
}
|