[PATCH V3 0/8] sysfs: drivers core: Add and use sysfs_emit and sysfs_emit_at

[PATCH V3 0/8] sysfs: drivers core: Add and use sysfs_emit and sysfs_emit_at

[Joe Perches]

Output defects can exist in sysfs content using sprintf and snprintf.

sprintf does not know the PAGE_SIZE maximum of the temporary buffer
used for outputting sysfs content and it's possible to overrun the
PAGE_SIZE buffer length.

Add a generic sysfs_emit function that knows that the size of the
temporary buffer and ensures that no overrun is done.

Add a generic sysfs_emit_at function that can be used in multiple
call situations that also ensures that no overrun is done.

V2: Add drivers core conversions
V3: Add a few more neatenings and conversions to drivers core
    Add mm hugetlb_report_node_meminfo conversion

Joe Perches (8):
  sysfs: Add sysfs_emit and sysfs_emit_at to format sysfs output
  drivers core: Use sysfs_emit and sysfs_emit_at for show(device *...) functions
  drivers core: Remove strcat uses around sysfs_emit and neaten
  drivers core: Reindent a couple uses around sysfs_emit
  drivers core: Miscellaneous changes for sysfs_emit
  mm: and drivers core: Convert hugetlb_report_node_meminfo to sysfs_emit
  drivers core: Use sysfs_emit for shared_cpu_map_show and shared_cpu_list_show
  drivers core: node: Use a more typical macro definition style for ACCESS_ATTR

 Documentation/filesystems/sysfs.rst     |   8 +-
 drivers/base/arch_topology.c            |   2 +-
 drivers/base/bus.c                      |   2 +-
 drivers/base/cacheinfo.c                |  49 ++--
 drivers/base/class.c                    |   2 +-
 drivers/base/core.c                     |  59 +++--
 drivers/base/cpu.c                      |  84 ++++---
 drivers/base/dd.c                       |   3 +-
 drivers/base/devcoredump.c              |   2 +-
 drivers/base/firmware_loader/fallback.c |   4 +-
 drivers/base/memory.c                   |  62 ++---
 drivers/base/node.c                     | 306 ++++++++++++------------
 drivers/base/platform.c                 |  17 +-
 drivers/base/power/sysfs.c              | 160 ++++++++-----
 drivers/base/power/wakeup_stats.c       |  17 +-
 drivers/base/soc.c                      |  64 +++--
 drivers/base/topology.c                 |  10 +-
 fs/sysfs/file.c                         |  55 +++++
 include/linux/hugetlb.h                 |   4 +-
 include/linux/sysfs.h                   |  15 ++
 mm/hugetlb.c                            |  18 +-
 21 files changed, 532 insertions(+), 411 deletions(-)

-- 
2.26.0

[PATCH V3 6/8] mm: and drivers core: Convert hugetlb_report_node_meminfo to sysfs_emit

[Joe Perches]

Convert the unbound sprintf in hugetlb_report_node_meminfo to use
sysfs_emit_at so that no possible overrun of a PAGE_SIZE buf can occur.

Signed-off-by: Joe Perches <joe@perches.com>
---
 drivers/base/node.c     |  2 +-
 include/linux/hugetlb.h |  4 ++--
 mm/hugetlb.c            | 18 ++++++++++--------
 3 files changed, 13 insertions(+), 11 deletions(-)

diff --git a/drivers/base/node.c b/drivers/base/node.c
index b33526a9fcfc..dafe03e82e7c 100644
--- a/drivers/base/node.c
+++ b/drivers/base/node.c
@@ -473,7 +473,7 @@ static ssize_t node_read_meminfo(struct device *dev,
 				    HPAGE_PMD_NR)
 #endif
 			    );
-	len += hugetlb_report_node_meminfo(nid, buf + len);
+	len += hugetlb_report_node_meminfo(buf, len, nid);
 	return len;
 }
 
diff --git a/include/linux/hugetlb.h b/include/linux/hugetlb.h
index d5cc5f802dd4..ebca2ef02212 100644
--- a/include/linux/hugetlb.h
+++ b/include/linux/hugetlb.h
@@ -129,7 +129,7 @@ void __unmap_hugepage_range(struct mmu_gather *tlb, struct vm_area_struct *vma,
 				unsigned long start, unsigned long end,
 				struct page *ref_page);
 void hugetlb_report_meminfo(struct seq_file *);
-int hugetlb_report_node_meminfo(int, char *);
+int hugetlb_report_node_meminfo(char *buf, int len, int nid);
 void hugetlb_show_meminfo(void);
 unsigned long hugetlb_total_pages(void);
 vm_fault_t hugetlb_fault(struct mm_struct *mm, struct vm_area_struct *vma,
@@ -245,7 +245,7 @@ static inline void hugetlb_report_meminfo(struct seq_file *m)
 {
 }
 
-static inline int hugetlb_report_node_meminfo(int nid, char *buf)
+static inline int hugetlb_report_node_meminfo(char *buf, int len, int nid)
 {
 	return 0;
 }
diff --git a/mm/hugetlb.c b/mm/hugetlb.c
index 61469fd3ad92..fe76f8fd5a73 100644
--- a/mm/hugetlb.c
+++ b/mm/hugetlb.c
@@ -3571,18 +3571,20 @@ void hugetlb_report_meminfo(struct seq_file *m)
 	seq_printf(m, "Hugetlb:        %8lu kB\n", total / 1024);
 }
 
-int hugetlb_report_node_meminfo(int nid, char *buf)
+int hugetlb_report_node_meminfo(char *buf, int len, int nid)
 {
 	struct hstate *h = &default_hstate;
+
 	if (!hugepages_supported())
 		return 0;
-	return sprintf(buf,
-		"Node %d HugePages_Total: %5u\n"
-		"Node %d HugePages_Free:  %5u\n"
-		"Node %d HugePages_Surp:  %5u\n",
-		nid, h->nr_huge_pages_node[nid],
-		nid, h->free_huge_pages_node[nid],
-		nid, h->surplus_huge_pages_node[nid]);
+
+	return sysfs_emit_at(buf, len,
+			     "Node %d HugePages_Total: %5u\n"
+			     "Node %d HugePages_Free:  %5u\n"
+			     "Node %d HugePages_Surp:  %5u\n",
+			     nid, h->nr_huge_pages_node[nid],
+			     nid, h->free_huge_pages_node[nid],
+			     nid, h->surplus_huge_pages_node[nid]);
 }
 
 void hugetlb_show_meminfo(void)
-- 
2.26.0

Re: [PATCH V3 6/8] mm: and drivers core: Convert hugetlb_report_node_meminfo to sysfs_emit

[Mike Kravetz]

On 9/16/20 1:40 PM, Joe Perches wrote:
> Convert the unbound sprintf in hugetlb_report_node_meminfo to use
> sysfs_emit_at so that no possible overrun of a PAGE_SIZE buf can occur.
> 
> Signed-off-by: Joe Perches <joe@perches.com>

Acked-by: Mike Kravetz <mike.kravetz@oracle.com>

-- 
Mike Kravetz

Re: [PATCH V3 6/8] mm: and drivers core: Convert hugetlb_report_node_meminfo to sysfs_emit

[Greg Kroah-Hartman]

On Wed, Sep 16, 2020 at 01:40:43PM -0700, Joe Perches wrote:
> Convert the unbound sprintf in hugetlb_report_node_meminfo to use
> sysfs_emit_at so that no possible overrun of a PAGE_SIZE buf can occur.
> 
> Signed-off-by: Joe Perches <joe@perches.com>
> ---
>  drivers/base/node.c     |  2 +-
>  include/linux/hugetlb.h |  4 ++--
>  mm/hugetlb.c            | 18 ++++++++++--------
>  3 files changed, 13 insertions(+), 11 deletions(-)
> 
> diff --git a/drivers/base/node.c b/drivers/base/node.c
> index b33526a9fcfc..dafe03e82e7c 100644
> --- a/drivers/base/node.c
> +++ b/drivers/base/node.c
> @@ -473,7 +473,7 @@ static ssize_t node_read_meminfo(struct device *dev,
>  				    HPAGE_PMD_NR)
>  #endif
>  			    );
> -	len += hugetlb_report_node_meminfo(nid, buf + len);
> +	len += hugetlb_report_node_meminfo(buf, len, nid);
>  	return len;
>  }
>  
> diff --git a/include/linux/hugetlb.h b/include/linux/hugetlb.h
> index d5cc5f802dd4..ebca2ef02212 100644
> --- a/include/linux/hugetlb.h
> +++ b/include/linux/hugetlb.h
> @@ -129,7 +129,7 @@ void __unmap_hugepage_range(struct mmu_gather *tlb, struct vm_area_struct *vma,
>  				unsigned long start, unsigned long end,
>  				struct page *ref_page);
>  void hugetlb_report_meminfo(struct seq_file *);
> -int hugetlb_report_node_meminfo(int, char *);
> +int hugetlb_report_node_meminfo(char *buf, int len, int nid);
>  void hugetlb_show_meminfo(void);
>  unsigned long hugetlb_total_pages(void);
>  vm_fault_t hugetlb_fault(struct mm_struct *mm, struct vm_area_struct *vma,
> @@ -245,7 +245,7 @@ static inline void hugetlb_report_meminfo(struct seq_file *m)
>  {
>  }
>  
> -static inline int hugetlb_report_node_meminfo(int nid, char *buf)
> +static inline int hugetlb_report_node_meminfo(char *buf, int len, int nid)
>  {
>  	return 0;
>  }
> diff --git a/mm/hugetlb.c b/mm/hugetlb.c
> index 61469fd3ad92..fe76f8fd5a73 100644
> --- a/mm/hugetlb.c
> +++ b/mm/hugetlb.c
> @@ -3571,18 +3571,20 @@ void hugetlb_report_meminfo(struct seq_file *m)
>  	seq_printf(m, "Hugetlb:        %8lu kB\n", total / 1024);
>  }
>  
> -int hugetlb_report_node_meminfo(int nid, char *buf)
> +int hugetlb_report_node_meminfo(char *buf, int len, int nid)
>  {
>  	struct hstate *h = &default_hstate;
> +
>  	if (!hugepages_supported())
>  		return 0;
> -	return sprintf(buf,
> -		"Node %d HugePages_Total: %5u\n"
> -		"Node %d HugePages_Free:  %5u\n"
> -		"Node %d HugePages_Surp:  %5u\n",
> -		nid, h->nr_huge_pages_node[nid],
> -		nid, h->free_huge_pages_node[nid],
> -		nid, h->surplus_huge_pages_node[nid]);
> +
> +	return sysfs_emit_at(buf, len,
> +			     "Node %d HugePages_Total: %5u\n"
> +			     "Node %d HugePages_Free:  %5u\n"
> +			     "Node %d HugePages_Surp:  %5u\n",
> +			     nid, h->nr_huge_pages_node[nid],
> +			     nid, h->free_huge_pages_node[nid],
> +			     nid, h->surplus_huge_pages_node[nid]);
>  }

That is NOT one-value-per-file, which is required for sysfs files.  This
should be 3 different sysfs files.

Ugh.

But that's separate from this series, thanks for redoing this.  I'll
take a look at it on Monday...

thanks,

greg k-h

Re: [PATCH V3 6/8] mm: and drivers core: Convert hugetlb_report_node_meminfo to sysfs_emit

[Joe Perches]

On Sat, 2020-09-19 at 08:22 +0200, Greg Kroah-Hartman wrote:
> I'll take a look at it on Monday...

It's decidedly not urgent/critical.

There are several logical defects all over the kernel
for these unbounded sprintfs, but I don't know of an
actual instance where the PAGE_SIZE buf is overfilled.

Re: [PATCH V3 6/8] mm: and drivers core: Convert hugetlb_report_node_meminfo to sysfs_emit

[Joe Perches]

On Sat, 2020-09-19 at 08:22 +0200, Greg Kroah-Hartman wrote:
> I'll take a look at it on Monday...

Thoughts?

Re: [PATCH V3 6/8] mm: and drivers core: Convert hugetlb_report_node_meminfo to sysfs_emit

[Joe Perches]

On Sat, 2020-09-19 at 08:22 +0200, Greg Kroah-Hartman wrote:
> On Wed, Sep 16, 2020 at 01:40:43PM -0700, Joe Perches wrote:
> > Convert the unbound sprintf in hugetlb_report_node_meminfo to use
> > sysfs_emit_at so that no possible overrun of a PAGE_SIZE buf can occur.
[]
> I'll take a look at it on Monday...

(noting that you didn't say _which_ Monday...)

Greg are you going to take this or should I ask someone else?

The first sysfs_emit addition patch was first posted about a month
ago without much change.

I think it'd be useful to get it into -next before 5.9 is released.

Re: [PATCH V3 6/8] mm: and drivers core: Convert hugetlb_report_node_meminfo to sysfs_emit

[Greg Kroah-Hartman]

On Mon, Sep 28, 2020 at 05:53:07PM -0700, Joe Perches wrote:
> On Sat, 2020-09-19 at 08:22 +0200, Greg Kroah-Hartman wrote:
> > On Wed, Sep 16, 2020 at 01:40:43PM -0700, Joe Perches wrote:
> > > Convert the unbound sprintf in hugetlb_report_node_meminfo to use
> > > sysfs_emit_at so that no possible overrun of a PAGE_SIZE buf can occur.
> []
> > I'll take a look at it on Monday...
> 
> (noting that you didn't say _which_ Monday...)

True :)

> Greg are you going to take this or should I ask someone else?
> 
> The first sysfs_emit addition patch was first posted about a month
> ago without much change.
> 
> I think it'd be useful to get it into -next before 5.9 is released.

I'll dig into this today...