[PATCH] uml: fault handler micro-cleanups
authorPaolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it>
Sat, 3 Sep 2005 22:57:26 +0000 (15:57 -0700)
committerLinus Torvalds <torvalds@evo.osdl.org>
Mon, 5 Sep 2005 07:06:21 +0000 (00:06 -0700)
Avoid chomping low bits of address for functions doing it by themselves,
fix whitespace, add a correctness checking.

I did this for remap-file-pages protection support, it was useful on its
own too.

Signed-off-by: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it>
Cc: Jeff Dike <jdike@addtoit.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
arch/um/kernel/trap_kern.c

index bef8abd5ea9ec6ce94b137f38f405325637366e3..b5fc89fe9eab8f2e51859e5a44c5f2bf5d226ab4 100644 (file)
@@ -26,6 +26,7 @@
 #include "mem.h"
 #include "mem_kern.h"
 
+/* Note this is constrained to return 0, -EFAULT, -EACCESS, -ENOMEM by segv(). */
 int handle_page_fault(unsigned long address, unsigned long ip, 
                      int is_write, int is_user, int *code_out)
 {
@@ -35,7 +36,6 @@ int handle_page_fault(unsigned long address, unsigned long ip,
        pud_t *pud;
        pmd_t *pmd;
        pte_t *pte;
-       unsigned long page;
        int err = -EFAULT;
 
        *code_out = SEGV_MAPERR;
@@ -52,7 +52,7 @@ int handle_page_fault(unsigned long address, unsigned long ip,
        else if(expand_stack(vma, address)) 
                goto out;
 
- good_area:
+good_area:
        *code_out = SEGV_ACCERR;
        if(is_write && !(vma->vm_flags & VM_WRITE)) 
                goto out;
@@ -60,9 +60,8 @@ int handle_page_fault(unsigned long address, unsigned long ip,
         if(!(vma->vm_flags & (VM_READ | VM_EXEC)))
                 goto out;
 
-       page = address & PAGE_MASK;
        do {
- survive:
+survive:
                switch (handle_mm_fault(mm, vma, address, is_write)){
                case VM_FAULT_MINOR:
                        current->min_flt++;
@@ -79,16 +78,16 @@ int handle_page_fault(unsigned long address, unsigned long ip,
                default:
                        BUG();
                }
-               pgd = pgd_offset(mm, page);
-               pud = pud_offset(pgd, page);
-               pmd = pmd_offset(pud, page);
-               pte = pte_offset_kernel(pmd, page);
+               pgd = pgd_offset(mm, address);
+               pud = pud_offset(pgd, address);
+               pmd = pmd_offset(pud, address);
+               pte = pte_offset_kernel(pmd, address);
        } while(!pte_present(*pte));
        err = 0;
        *pte = pte_mkyoung(*pte);
        if(pte_write(*pte)) *pte = pte_mkdirty(*pte);
-       flush_tlb_page(vma, page);
- out:
+       flush_tlb_page(vma, address);
+out:
        up_read(&mm->mmap_sem);
        return(err);
 
@@ -144,19 +143,18 @@ unsigned long segv(struct faultinfo fi, unsigned long ip, int is_user, void *sc)
                panic("Kernel mode fault at addr 0x%lx, ip 0x%lx", 
                      address, ip);
 
-       if(err == -EACCES){
+       if (err == -EACCES) {
                si.si_signo = SIGBUS;
                si.si_errno = 0;
                si.si_code = BUS_ADRERR;
                si.si_addr = (void *)address;
                 current->thread.arch.faultinfo = fi;
                force_sig_info(SIGBUS, &si, current);
-       }
-       else if(err == -ENOMEM){
+       } else if (err == -ENOMEM) {
                printk("VM: killing process %s\n", current->comm);
                do_exit(SIGKILL);
-       }
-       else {
+       } else {
+               BUG_ON(err != -EFAULT);
                si.si_signo = SIGSEGV;
                si.si_addr = (void *) address;
                 current->thread.arch.faultinfo = fi;