Re: [RFC 1/3] mmiotrace full patch, preview 3

[Pekka Paalanen <pq@iki.fi>]

On Mon, 28 Apr 2008 21:45:04 +0300
Pekka Paalanen <pq@iki.fi> wrote:

> Hi all,
> 
> this is the third full mmiotrace patch set for review (a bit late, sorry).
> These patches do not apply to any git tree that I know of, but after Ingo
> pushes a new version of his sched-devel/latest tree, these patches should
> be subtractable from there.
...
> This first patch adds the bulk of the new code, only introducing new files.
> The second patch establishes the interfaces to the ftrace framework, and
> the third patch hooks everything up into the Kbuild system, page fault
> handler, and ioremap functions.

 arch/x86/Kconfig.debug |   28 ++++++++++++++++++++++++++++
 arch/x86/mm/Makefile   |    5 +++++
 arch/x86/mm/fault.c    |   13 +++++++++++++
 arch/x86/mm/ioremap.c  |    9 ++++++++-
 arch/x86/mm/pageattr.c |    1 +
 5 files changed, 55 insertions(+), 1 deletions(-)

diff --git a/arch/x86/Kconfig.debug b/arch/x86/Kconfig.debug
index 1f237d2..fb0794e 100644
--- a/arch/x86/Kconfig.debug
+++ b/arch/x86/Kconfig.debug
@@ -170,6 +170,34 @@ config IOMMU_LEAK
 	  Add a simple leak tracer to the IOMMU code. This is useful when you
 	  are debugging a buggy device driver that leaks IOMMU mappings.
 
+config MMIOTRACE_HOOKS
+	bool
+
+config MMIOTRACE
+	bool "Memory mapped IO tracing"
+	depends on DEBUG_KERNEL && PCI
+	select TRACING
+	select MMIOTRACE_HOOKS
+	default y
+	help
+	  Mmiotrace traces Memory Mapped I/O access and is meant for
+	  debugging and reverse engineering. It is called from the ioremap
+	  implementation and works via page faults. Tracing is disabled by
+	  default and can be enabled at run-time.
+
+	  See Documentation/tracers/mmiotrace.txt.
+	  If you are not helping to develop drivers, say N.
+
+config MMIOTRACE_TEST
+	tristate "Test module for mmiotrace"
+	depends on MMIOTRACE && m
+	help
+	  This is a dumb module for testing mmiotrace. It is very dangerous
+	  as it will write garbage to IO memory starting at a given address.
+	  However, it should be safe to use on e.g. unused portion of VRAM.
+
+	  Say N, unless you absolutely know what you are doing.
+
 #
 # IO delay types:
 #
diff --git a/arch/x86/mm/Makefile b/arch/x86/mm/Makefile
index b7b3e4c..07dab50 100644
--- a/arch/x86/mm/Makefile
+++ b/arch/x86/mm/Makefile
@@ -8,6 +8,11 @@ obj-$(CONFIG_X86_PTDUMP)	+= dump_pagetables.o
 
 obj-$(CONFIG_HIGHMEM)		+= highmem_32.o
 
+obj-$(CONFIG_MMIOTRACE_HOOKS)	+= kmmio.o
+obj-$(CONFIG_MMIOTRACE)		+= mmiotrace.o
+mmiotrace-y			:= pf_in.o mmio-mod.o
+obj-$(CONFIG_MMIOTRACE_TEST)	+= testmmiotrace.o
+
 ifeq ($(CONFIG_X86_32),y)
 obj-$(CONFIG_NUMA)		+= discontig_32.o
 else
diff --git a/arch/x86/mm/fault.c b/arch/x86/mm/fault.c
index 4145597..0440dce 100644
--- a/arch/x86/mm/fault.c
+++ b/arch/x86/mm/fault.c
@@ -26,6 +26,7 @@
 #include <linux/uaccess.h>
 #include <linux/kdebug.h>
 #include <linux/magic.h>
+#include <linux/mmiotrace.h>
 
 #include <asm/system.h>
 #include <asm/desc.h>
@@ -51,6 +52,16 @@
 #define PF_RSVD		(1<<3)
 #define PF_INSTR	(1<<4)
 
+static inline int kmmio_fault(struct pt_regs *regs, unsigned long addr)
+{
+#ifdef CONFIG_MMIOTRACE_HOOKS
+	if (unlikely(is_kmmio_active()))
+		if (kmmio_handler(regs, addr) == 1)
+			return -1;
+#endif
+	return 0;
+}
+
 static inline int notify_page_fault(struct pt_regs *regs)
 {
 #ifdef CONFIG_KPROBES
@@ -623,6 +634,8 @@ void __kprobes do_page_fault(struct pt_regs *regs, unsigned long error_code)
 
 	if (notify_page_fault(regs))
 		return;
+	if (unlikely(kmmio_fault(regs, address)))
+		return;
 
 	/*
 	 * We fault-in kernel-space virtual memory on-demand. The
diff --git a/arch/x86/mm/ioremap.c b/arch/x86/mm/ioremap.c
index d176b23..68b287d 100644
--- a/arch/x86/mm/ioremap.c
+++ b/arch/x86/mm/ioremap.c
@@ -12,6 +12,7 @@
 #include <linux/module.h>
 #include <linux/slab.h>
 #include <linux/vmalloc.h>
+#include <linux/mmiotrace.h>
 
 #include <asm/cacheflush.h>
 #include <asm/e820.h>
@@ -126,6 +127,7 @@ static void __iomem *__ioremap(resource_size_t phys_addr, unsigned long size,
 	unsigned long new_prot_val;
 	pgprot_t prot;
 	int retval;
+	void __iomem *ret_addr;
 
 	/* Don't allow wraparound or zero size */
 	last_addr = phys_addr + size - 1;
@@ -229,7 +231,10 @@ static void __iomem *__ioremap(resource_size_t phys_addr, unsigned long size,
 		return NULL;
 	}
 
-	return (void __iomem *) (vaddr + offset);
+	ret_addr = (void __iomem *) (vaddr + offset);
+	mmiotrace_ioremap(phys_addr, size, ret_addr);
+
+	return ret_addr;
 }
 
 /**
@@ -306,6 +311,8 @@ void iounmap(volatile void __iomem *addr)
 	    addr < phys_to_virt(ISA_END_ADDRESS))
 		return;
 
+	mmiotrace_iounmap(addr);
+
 	addr = (volatile void __iomem *)
 		(PAGE_MASK & (unsigned long __force)addr);
 
diff --git a/arch/x86/mm/pageattr.c b/arch/x86/mm/pageattr.c
index bd5e05c..099374b 100644
--- a/arch/x86/mm/pageattr.c
+++ b/arch/x86/mm/pageattr.c
@@ -227,6 +227,7 @@ pte_t *lookup_address(unsigned long address, unsigned int *level)
 
 	return pte_offset_kernel(pmd, address);
 }
+EXPORT_SYMBOL_GPL(lookup_address);
 
 /*
  * Set the new pmd in all the pgds we know about: