[PATCH] powerpc vdso updates

STALE7359d

5 messages, 3 authors, 2006-06-10 · open the first message on its own page

[PATCH] powerpc vdso updates

From: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Date: 2006-05-30 03:52:05

This patch cleans up some locking & error handling in the ppc vdso and
moves the vdso base pointer from the thread struct to the mm context
where it more logically belongs. It brings the powerpc implementation
closer to Ingo's new x86 one and also adds an arch_vma_name() function
allowing to print [vsdo] in /proc/<pid>/maps if Ingo's x86 vdso patch is
also applied.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
---

This is 2.6.18 material, hopefully should go along with Ingo's x86 vdso updates. Ingo, if you
change something to arch_vma_name(), please let me know.



Index: linux-work/arch/powerpc/kernel/signal_32.c
===================================================================
--- linux-work.orig/arch/powerpc/kernel/signal_32.c	2006-05-30 13:34:09.000000000 +1000
+++ linux-work/arch/powerpc/kernel/signal_32.c	2006-05-30 13:40:21.000000000 +1000
@@ -757,10 +757,10 @@ static int handle_rt_signal(unsigned lon
 
 	/* Save user registers on the stack */
 	frame = &rt_sf->uc.uc_mcontext;
-	if (vdso32_rt_sigtramp && current->thread.vdso_base) {
+	if (vdso32_rt_sigtramp && current->mm->context.vdso_base) {
 		if (save_user_regs(regs, frame, 0))
 			goto badframe;
-		regs->link = current->thread.vdso_base + vdso32_rt_sigtramp;
+		regs->link = current->mm->context.vdso_base + vdso32_rt_sigtramp;
 	} else {
 		if (save_user_regs(regs, frame, __NR_rt_sigreturn))
 			goto badframe;
@@ -1029,10 +1029,10 @@ static int handle_signal(unsigned long s
 	    || __put_user(sig, &sc->signal))
 		goto badframe;
 
-	if (vdso32_sigtramp && current->thread.vdso_base) {
+	if (vdso32_sigtramp && current->mm->context.vdso_base) {
 		if (save_user_regs(regs, &frame->mctx, 0))
 			goto badframe;
-		regs->link = current->thread.vdso_base + vdso32_sigtramp;
+		regs->link = current->mm->context.vdso_base + vdso32_sigtramp;
 	} else {
 		if (save_user_regs(regs, &frame->mctx, __NR_sigreturn))
 			goto badframe;
Index: linux-work/arch/powerpc/kernel/signal_64.c
===================================================================
--- linux-work.orig/arch/powerpc/kernel/signal_64.c	2006-05-30 13:34:09.000000000 +1000
+++ linux-work/arch/powerpc/kernel/signal_64.c	2006-05-30 13:40:21.000000000 +1000
@@ -394,8 +394,8 @@ static int setup_rt_frame(int signr, str
 	current->thread.fpscr.val = 0;
 
 	/* Set up to return from userspace. */
-	if (vdso64_rt_sigtramp && current->thread.vdso_base) {
-		regs->link = current->thread.vdso_base + vdso64_rt_sigtramp;
+	if (vdso64_rt_sigtramp && current->mm->context.vdso_base) {
+		regs->link = current->mm->context.vdso_base + vdso64_rt_sigtramp;
 	} else {
 		err |= setup_trampoline(__NR_rt_sigreturn, &frame->tramp[0]);
 		if (err)
Index: linux-work/arch/powerpc/kernel/vdso.c
===================================================================
--- linux-work.orig/arch/powerpc/kernel/vdso.c	2006-05-30 13:34:09.000000000 +1000
+++ linux-work/arch/powerpc/kernel/vdso.c	2006-05-30 13:41:31.000000000 +1000
@@ -223,6 +223,7 @@ int arch_setup_additional_pages(struct l
 	struct vm_area_struct *vma;
 	unsigned long vdso_pages;
 	unsigned long vdso_base;
+	int rc;
 
 #ifdef CONFIG_PPC64
 	if (test_thread_flag(TIF_32BIT)) {
@@ -237,20 +238,13 @@ int arch_setup_additional_pages(struct l
 	vdso_base = VDSO32_MBASE;
 #endif
 
-	current->thread.vdso_base = 0;
+	current->mm->context.vdso_base = 0;
 
 	/* vDSO has a problem and was disabled, just don't "enable" it for the
 	 * process
 	 */
 	if (vdso_pages == 0)
 		return 0;
-
-	vma = kmem_cache_alloc(vm_area_cachep, SLAB_KERNEL);
-	if (vma == NULL)
-		return -ENOMEM;
-
-	memset(vma, 0, sizeof(*vma));
-
 	/* Add a page to the vdso size for the data page */
 	vdso_pages ++;
 
@@ -259,17 +253,23 @@ int arch_setup_additional_pages(struct l
 	 * at vdso_base which is the "natural" base for it, but we might fail
 	 * and end up putting it elsewhere.
 	 */
+	down_write(&mm->mmap_sem);
 	vdso_base = get_unmapped_area(NULL, vdso_base,
 				      vdso_pages << PAGE_SHIFT, 0, 0);
-	if (vdso_base & ~PAGE_MASK) {
-		kmem_cache_free(vm_area_cachep, vma);
-		return (int)vdso_base;
+	if (IS_ERR_VALUE(vdso_base)) {
+		rc = vdso_base;
+		goto fail_mmapsem;
 	}
 
-	current->thread.vdso_base = vdso_base;
 
+	/* Allocate a VMA structure and fill it up */
+	vma = kmem_cache_zalloc(vm_area_cachep, SLAB_KERNEL);
+	if (vma == NULL) {
+		rc = -ENOMEM;
+		goto fail_mmapsem;
+	}
 	vma->vm_mm = mm;
-	vma->vm_start = current->thread.vdso_base;
+	vma->vm_start = vdso_base;
 	vma->vm_end = vma->vm_start + (vdso_pages << PAGE_SHIFT);
 
 	/*
@@ -282,23 +282,38 @@ int arch_setup_additional_pages(struct l
 	 * It's fine to use that for setting breakpoints in the vDSO code
 	 * pages though
 	 */
-	vma->vm_flags = VM_READ | VM_EXEC | VM_MAYREAD | VM_MAYWRITE | VM_MAYEXEC;
+	vma->vm_flags = VM_READ|VM_EXEC|VM_MAYREAD|VM_MAYWRITE|VM_MAYEXEC;
 	vma->vm_flags |= mm->def_flags;
 	vma->vm_page_prot = protection_map[vma->vm_flags & 0x7];
 	vma->vm_ops = &vdso_vmops;
 
-	down_write(&mm->mmap_sem);
-	if (insert_vm_struct(mm, vma)) {
-		up_write(&mm->mmap_sem);
-		kmem_cache_free(vm_area_cachep, vma);
-		return -ENOMEM;
-	}
+	/* Insert new VMA */
+	rc = insert_vm_struct(mm, vma);
+	if (rc)
+		goto fail_vma;
+
+	/* Put vDSO base into mm struct and account for memory usage */
+	current->mm->context.vdso_base = vdso_base;
 	mm->total_vm += (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
 	up_write(&mm->mmap_sem);
-
 	return 0;
+
+ fail_vma:
+	kmem_cache_free(vm_area_cachep, vma);
+ fail_mmapsem:
+	up_write(&mm->mmap_sem);
+	return rc;
+}
+
+const char *arch_vma_name(struct vm_area_struct *vma)
+{
+	if (vma->vm_mm && vma->vm_start == vma->vm_mm->context.vdso_base)
+		return "[vdso]";
+	return NULL;
 }
 
+
+
 static void * __init find_section32(Elf32_Ehdr *ehdr, const char *secname,
 				  unsigned long *size)
 {
Index: linux-work/include/asm-powerpc/elf.h
===================================================================
--- linux-work.orig/include/asm-powerpc/elf.h	2006-05-30 13:34:09.000000000 +1000
+++ linux-work/include/asm-powerpc/elf.h	2006-05-30 13:40:21.000000000 +1000
@@ -294,7 +294,7 @@ do {									\
 	NEW_AUX_ENT(AT_DCACHEBSIZE, dcache_bsize);			\
 	NEW_AUX_ENT(AT_ICACHEBSIZE, icache_bsize);			\
 	NEW_AUX_ENT(AT_UCACHEBSIZE, ucache_bsize);			\
-	VDSO_AUX_ENT(AT_SYSINFO_EHDR, current->thread.vdso_base)	\
+	VDSO_AUX_ENT(AT_SYSINFO_EHDR, current->mm->context.vdso_base)	\
 } while (0)
 
 /* PowerPC64 relocations defined by the ABIs */
Index: linux-work/include/asm-powerpc/mmu.h
===================================================================
--- linux-work.orig/include/asm-powerpc/mmu.h	2006-05-30 13:34:09.000000000 +1000
+++ linux-work/include/asm-powerpc/mmu.h	2006-05-30 13:40:21.000000000 +1000
@@ -360,6 +360,7 @@ typedef struct {
 #ifdef CONFIG_HUGETLB_PAGE
 	u16 low_htlb_areas, high_htlb_areas;
 #endif
+	unsigned long vdso_base;
 } mm_context_t;
 
 
Index: linux-work/include/asm-powerpc/page.h
===================================================================
--- linux-work.orig/include/asm-powerpc/page.h	2006-05-30 13:34:09.000000000 +1000
+++ linux-work/include/asm-powerpc/page.h	2006-05-30 13:40:21.000000000 +1000
@@ -198,6 +198,9 @@ extern void copy_user_page(void *to, voi
 		struct page *p);
 extern int page_is_ram(unsigned long pfn);
 
+struct vm_area_struct;
+extern const char *arch_vma_name(struct vm_area_struct *vma);
+
 #include <asm-generic/memory_model.h>
 #endif /* __ASSEMBLY__ */
 
Index: linux-work/include/asm-powerpc/processor.h
===================================================================
--- linux-work.orig/include/asm-powerpc/processor.h	2006-05-30 13:34:09.000000000 +1000
+++ linux-work/include/asm-powerpc/processor.h	2006-05-30 13:40:21.000000000 +1000
@@ -153,7 +153,6 @@ struct thread_struct {
 	unsigned long	start_tb;	/* Start purr when proc switched in */
 	unsigned long	accum_tb;	/* Total accumilated purr for process */
 #endif
-	unsigned long	vdso_base;	/* base of the vDSO library */
 	unsigned long	dabr;		/* Data address breakpoint register */
 #ifdef CONFIG_ALTIVEC
 	/* Complete AltiVec register set */

Re: [PATCH] powerpc vdso updates

From: Ingo Molnar <hidden>
Date: 2006-05-30 06:24:31

* Benjamin Herrenschmidt [off-list ref] wrote:
This patch cleans up some locking & error handling in the ppc vdso and 
moves the vdso base pointer from the thread struct to the mm context 
where it more logically belongs. It brings the powerpc implementation 
closer to Ingo's new x86 one and also adds an arch_vma_name() function 
allowing to print [vsdo] in /proc/<pid>/maps if Ingo's x86 vdso patch 
is also applied.
looks good to me.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Reviewed-by: Ingo Molnar <redacted>
This is 2.6.18 material, hopefully should go along with Ingo's x86 
vdso updates. Ingo, if you change something to arch_vma_name(), please 
let me know.
ok. There's one bit that could potentially be 2.6.17 material:
 	 * at vdso_base which is the "natural" base for it, but we might fail
 	 * and end up putting it elsewhere.
 	 */
+	down_write(&mm->mmap_sem);
 	vdso_base = get_unmapped_area(NULL, vdso_base,
 				      vdso_pages << PAGE_SHIFT, 0, 0);
get_unmapped_area() without holding the mmap semaphore seems dangerous. 
The VDSO setup itself should be 'private' to the process, but i'm not 
totally sure that no other kernel code could get access to this mm. For 
example the swapout code? Am i missing something?

	Ingo

Re: [PATCH] powerpc vdso updates

From: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Date: 2006-05-30 07:22:08

get_unmapped_area() without holding the mmap semaphore seems dangerous. 
The VDSO setup itself should be 'private' to the process, but i'm not 
totally sure that no other kernel code could get access to this mm. For 
example the swapout code? Am i missing something?
Well, all of this happens so early, I doubt anything will cause actual
harm, for get_unmapped_area to be a problem, something would have to
concurrently allocate a VMA (or dispose of) I think that can't happen at
this point before we even run the binary but yes, better safe than
sorry... I supose the whole patch could get in 2.6.17 even without the
arch_vma_name (that is it can be there, it wont be used though until
your patch gets in).

Ben.

Re: [PATCH] powerpc vdso updates

From: Anton Blanchard <hidden>
Date: 2006-06-10 03:15:43

Hi Ben,
This patch cleans up some locking & error handling in the ppc vdso and
moves the vdso base pointer from the thread struct to the mm context
where it more logically belongs. It brings the powerpc implementation
closer to Ingo's new x86 one and also adds an arch_vma_name() function
allowing to print [vsdo] in /proc/<pid>/maps if Ingo's x86 vdso patch is
also applied.
Im getting this on the current powerpc git tree:

arch/powerpc/kernel/signal_32.c: In function 'handle_rt_signal':
arch/powerpc/kernel/signal_32.c:763: error: request for member
'vdso_base' in something not a structure or union

It looks like we use the mm_context_t definition in asm-ppc/mmu.h for
32bit ARCH=powerpc builds and need vdso_base there too.

Anton

Re: [PATCH] powerpc vdso updates

From: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Date: 2006-06-10 06:30:37

On Sat, 2006-06-10 at 13:15 +1000, Anton Blanchard wrote:
Hi Ben,
quoted
This patch cleans up some locking & error handling in the ppc vdso and
moves the vdso base pointer from the thread struct to the mm context
where it more logically belongs. It brings the powerpc implementation
closer to Ingo's new x86 one and also adds an arch_vma_name() function
allowing to print [vsdo] in /proc/<pid>/maps if Ingo's x86 vdso patch is
also applied.
Im getting this on the current powerpc git tree:

arch/powerpc/kernel/signal_32.c: In function 'handle_rt_signal':
arch/powerpc/kernel/signal_32.c:763: error: request for member
'vdso_base' in something not a structure or union

It looks like we use the mm_context_t definition in asm-ppc/mmu.h for
32bit ARCH=powerpc builds and need vdso_base there too.
Yes, I have a patch almost ready for that.

Ben.
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help