[v2 PATCH] cpufreq:stats: Handle the case when trans_table goes beyond PAGE_SIZE

[v2 PATCH] cpufreq:stats: Handle the case when trans_table goes beyond PAGE_SIZE

[Gautham R. Shenoy]

From: "Gautham R. Shenoy" <ego@linux.vnet.ibm.com>

On platforms with large number of Pstates, the transition table, which
is a NxN matrix, can overflow beyond the PAGE_SIZE boundary.

This can be seen on POWER9 which has 100+ Pstates.

As a result, each time the trans_table is read for any of the CPUs, we
will get the following error.

---------------------------------------------------
fill_read_buffer: show+0x0/0xa0 returned bad count
---------------------------------------------------

This patch ensures that in case of an overflow, we print a warning
once in the dmesg and return FILE TOO LARGE error for this and all
subsequent accesses of trans_table.

Signed-off-by: Gautham R. Shenoy <ego@linux.vnet.ibm.com>
---
 Documentation/cpu-freq/cpufreq-stats.txt | 3 +++
 drivers/cpufreq/cpufreq_stats.c          | 7 +++++--
 2 files changed, 8 insertions(+), 2 deletions(-)

diff --git a/Documentation/cpu-freq/cpufreq-stats.txt b/Documentation/cpu-freq/cpufreq-stats.txt
index 2bbe207..a873855 100644
--- a/Documentation/cpu-freq/cpufreq-stats.txt
+++ b/Documentation/cpu-freq/cpufreq-stats.txt
@@ -90,6 +90,9 @@ Freq_i to Freq_j. Freq_i is in descending order with increasing rows and
 Freq_j is in descending order with increasing columns. The output here also 
 contains the actual freq values for each row and column for better readability.
 
+If the transition table is bigger than PAGE_SIZE, reading this will
+return an -EFBIG error.
+
 --------------------------------------------------------------------------------
 <mysystem>:/sys/devices/system/cpu/cpu0/cpufreq/stats # cat trans_table
    From  :    To
diff --git a/drivers/cpufreq/cpufreq_stats.c b/drivers/cpufreq/cpufreq_stats.c
index e75880e..1e55b57 100644
--- a/drivers/cpufreq/cpufreq_stats.c
+++ b/drivers/cpufreq/cpufreq_stats.c
@@ -118,8 +118,11 @@ static ssize_t show_trans_table(struct cpufreq_policy *policy, char *buf)
 			break;
 		len += snprintf(buf + len, PAGE_SIZE - len, "\n");
 	}
-	if (len >= PAGE_SIZE)
-		return PAGE_SIZE;
+
+	if (len >= PAGE_SIZE) {
+		pr_warn_once("cpufreq transition table exceeds PAGE_SIZE. Disabling\n");
+		return -EFBIG;
+	}
 	return len;
 }
 cpufreq_freq_attr_ro(trans_table);
-- 
1.9.4

Re: [v2 PATCH] cpufreq:stats: Handle the case when trans_table goes beyond PAGE_SIZE

[Viresh Kumar]

On 07-11-17, 13:39, Gautham R. Shenoy wrote:
> From: "Gautham R. Shenoy" <ego@linux.vnet.ibm.com>
> 
> On platforms with large number of Pstates, the transition table, which
> is a NxN matrix, can overflow beyond the PAGE_SIZE boundary.
> 
> This can be seen on POWER9 which has 100+ Pstates.
> 
> As a result, each time the trans_table is read for any of the CPUs, we
> will get the following error.
> 
> ---------------------------------------------------
> fill_read_buffer: show+0x0/0xa0 returned bad count
> ---------------------------------------------------
> 
> This patch ensures that in case of an overflow, we print a warning
> once in the dmesg and return FILE TOO LARGE error for this and all
> subsequent accesses of trans_table.
> 
> Signed-off-by: Gautham R. Shenoy <ego@linux.vnet.ibm.com>
> ---
>  Documentation/cpu-freq/cpufreq-stats.txt | 3 +++
>  drivers/cpufreq/cpufreq_stats.c          | 7 +++++--
>  2 files changed, 8 insertions(+), 2 deletions(-)

Acked-by: Viresh Kumar <viresh.kumar@linaro.org>

-- 
viresh