2023-12-04 08:58:17

by Arnd Bergmann

[permalink] [raw]
Subject: [PATCH] [v2] net: hns3: reduce stack usage in hclge_dbg_dump_tm_pri()

From: Arnd Bergmann <[email protected]>

This function exceeds the stack frame warning limit:

drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_debugfs.c: In function 'hclge_dbg_dump_tm_pri':
drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_debugfs.c:1039:1: error: the frame size of 1408 bytes is larger than 1024 bytes [-Werror=frame-larger-than=]

Use dynamic allocation for the largest stack object instead. It
would be nice to rewrite this file to completely avoid the extra
buffer and just use the one that was already allocated by debugfs,
but that is a much larger change.

Signed-off-by: Arnd Bergmann <[email protected]>
---
v2: fix error handling leak
---
.../hisilicon/hns3/hns3pf/hclge_debugfs.c | 21 ++++++++++++-------
1 file changed, 14 insertions(+), 7 deletions(-)

diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_debugfs.c b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_debugfs.c
index ff3f8f424ad9..8f94e13c1edf 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_debugfs.c
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_debugfs.c
@@ -981,7 +981,7 @@ static const struct hclge_dbg_item tm_pri_items[] = {

static int hclge_dbg_dump_tm_pri(struct hclge_dev *hdev, char *buf, int len)
{
- char data_str[ARRAY_SIZE(tm_pri_items)][HCLGE_DBG_DATA_STR_LEN];
+ char *data_str;
struct hclge_tm_shaper_para c_shaper_para, p_shaper_para;
char *result[ARRAY_SIZE(tm_pri_items)], *sch_mode_str;
char content[HCLGE_DBG_TM_INFO_LEN];
@@ -992,8 +992,13 @@ static int hclge_dbg_dump_tm_pri(struct hclge_dev *hdev, char *buf, int len)
if (ret)
return ret;

+ data_str = kcalloc(ARRAY_SIZE(tm_pri_items), HCLGE_DBG_DATA_STR_LEN,
+ GFP_KERNEL);
+ if (!data_str)
+ return -ENOMEM;
+
for (i = 0; i < ARRAY_SIZE(tm_pri_items); i++)
- result[i] = &data_str[i][0];
+ result[i] = &data_str[i * HCLGE_DBG_DATA_STR_LEN];

hclge_dbg_fill_content(content, sizeof(content), tm_pri_items,
NULL, ARRAY_SIZE(tm_pri_items));
@@ -1002,23 +1007,23 @@ static int hclge_dbg_dump_tm_pri(struct hclge_dev *hdev, char *buf, int len)
for (i = 0; i < pri_num; i++) {
ret = hclge_tm_get_pri_sch_mode(hdev, i, &sch_mode);
if (ret)
- return ret;
+ goto out;

ret = hclge_tm_get_pri_weight(hdev, i, &weight);
if (ret)
- return ret;
+ goto out;

ret = hclge_tm_get_pri_shaper(hdev, i,
HCLGE_OPC_TM_PRI_C_SHAPPING,
&c_shaper_para);
if (ret)
- return ret;
+ goto out;

ret = hclge_tm_get_pri_shaper(hdev, i,
HCLGE_OPC_TM_PRI_P_SHAPPING,
&p_shaper_para);
if (ret)
- return ret;
+ goto out;

sch_mode_str = sch_mode & HCLGE_TM_TX_SCHD_DWRR_MSK ? "dwrr" :
"sp";
@@ -1035,7 +1040,9 @@ static int hclge_dbg_dump_tm_pri(struct hclge_dev *hdev, char *buf, int len)
pos += scnprintf(buf + pos, len - pos, "%s", content);
}

- return 0;
+out:
+ kfree(data_str);
+ return ret;
}

static const struct hclge_dbg_item tm_qset_items[] = {
--
2.39.2


2023-12-04 14:51:09

by Jijie Shao

[permalink] [raw]
Subject: Re: [PATCH] [v2] net: hns3: reduce stack usage in hclge_dbg_dump_tm_pri()


on 2023/12/4 16:57, Arnd Bergmann wrote:
> s already allocated by debugfs,
> but that is a much larger change.
>
> Signed-off-by: Arnd Bergmann <[email protected]>
> ---
> v2: fix error handling leak
> ---
> .../hisilicon/hns3/hns3pf/hclge_debugfs.c | 21 ++++++++++++-------
> 1 file changed, 14 insertions(+), 7 deletions(-)
>
> diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_debugfs.c b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_debugfs.c
> index ff3f8f424ad9..8f94e13c1edf 100644
> --- a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_debugfs.c
> +++ b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_debugfs.c
> @@ -981,7 +981,7 @@ static const struct hclge_dbg_item tm_pri_items[] = {
>
> static int hclge_dbg_dump_tm_pri(struct hclge_dev *hdev, char *buf, int len)
> {
> - char data_str[ARRAY_SIZE(tm_pri_items)][HCLGE_DBG_DATA_STR_LEN];
> + char *data_str;

We want to define variables in an inverted triangle based on the code length.
so, "char *data_str" should move four lines down.

struct hclge_tm_shaper_para c_shaper_para, p_shaper_para;
char *result[ARRAY_SIZE(tm_pri_items)], *sch_mode_str;
char content[HCLGE_DBG_TM_INFO_LEN];
u8 pri_num, sch_mode, weight, i, j;
char *data_str;
int pos, ret;

> struct hclge_tm_shaper_para c_shaper_para, p_shaper_para;
> char *result[ARRAY_SIZE(tm_pri_items)], *sch_mode_str;
> char content[HCLGE_DBG_TM_INFO_LEN];
> @@ -992,8 +992,13 @@ static int hclge_dbg_dump_tm_pri(struct hclge_dev *hdev, char *buf, int len)
> if (ret)
> return ret;
>

2023-12-04 18:56:02

by Christophe JAILLET

[permalink] [raw]
Subject: Re: [PATCH] [v2] net: hns3: reduce stack usage in hclge_dbg_dump_tm_pri()

Le 04/12/2023 à 09:57, Arnd Bergmann a écrit :
> From: Arnd Bergmann <[email protected]>
>
> This function exceeds the stack frame warning limit:
>
> drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_debugfs.c: In function 'hclge_dbg_dump_tm_pri':
> drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_debugfs.c:1039:1: error: the frame size of 1408 bytes is larger than 1024 bytes [-Werror=frame-larger-than=]
>
> Use dynamic allocation for the largest stack object instead. It
> would be nice to rewrite this file to completely avoid the extra
> buffer and just use the one that was already allocated by debugfs,
> but that is a much larger change.
>
> Signed-off-by: Arnd Bergmann <[email protected]>
> ---
> v2: fix error handling leak
> ---
> .../hisilicon/hns3/hns3pf/hclge_debugfs.c | 21 ++++++++++++-------
> 1 file changed, 14 insertions(+), 7 deletions(-)
>
> diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_debugfs.c b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_debugfs.c
> index ff3f8f424ad9..8f94e13c1edf 100644
> --- a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_debugfs.c
> +++ b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_debugfs.c
> @@ -981,7 +981,7 @@ static const struct hclge_dbg_item tm_pri_items[] = {
>
> static int hclge_dbg_dump_tm_pri(struct hclge_dev *hdev, char *buf, int len)
> {
> - char data_str[ARRAY_SIZE(tm_pri_items)][HCLGE_DBG_DATA_STR_LEN];
> + char *data_str;
> struct hclge_tm_shaper_para c_shaper_para, p_shaper_para;
> char *result[ARRAY_SIZE(tm_pri_items)], *sch_mode_str;
> char content[HCLGE_DBG_TM_INFO_LEN];
> @@ -992,8 +992,13 @@ static int hclge_dbg_dump_tm_pri(struct hclge_dev *hdev, char *buf, int len)
> if (ret)
> return ret;
>
> + data_str = kcalloc(ARRAY_SIZE(tm_pri_items), HCLGE_DBG_DATA_STR_LEN,
> + GFP_KERNEL);
> + if (!data_str)
> + return -ENOMEM;
> +
> for (i = 0; i < ARRAY_SIZE(tm_pri_items); i++)
> - result[i] = &data_str[i][0];
> + result[i] = &data_str[i * HCLGE_DBG_DATA_STR_LEN];
>
> hclge_dbg_fill_content(content, sizeof(content), tm_pri_items,
> NULL, ARRAY_SIZE(tm_pri_items));
> @@ -1002,23 +1007,23 @@ static int hclge_dbg_dump_tm_pri(struct hclge_dev *hdev, char *buf, int len)
> for (i = 0; i < pri_num; i++) {
> ret = hclge_tm_get_pri_sch_mode(hdev, i, &sch_mode);
> if (ret)
> - return ret;
> + goto out;
>
> ret = hclge_tm_get_pri_weight(hdev, i, &weight);
> if (ret)
> - return ret;
> + goto out;
>
> ret = hclge_tm_get_pri_shaper(hdev, i,
> HCLGE_OPC_TM_PRI_C_SHAPPING,
> &c_shaper_para);
> if (ret)
> - return ret;
> + goto out;
>
> ret = hclge_tm_get_pri_shaper(hdev, i,
> HCLGE_OPC_TM_PRI_P_SHAPPING,
> &p_shaper_para);
> if (ret)
> - return ret;
> + goto out;
>
> sch_mode_str = sch_mode & HCLGE_TM_TX_SCHD_DWRR_MSK ? "dwrr" :
> "sp";
> @@ -1035,7 +1040,9 @@ static int hclge_dbg_dump_tm_pri(struct hclge_dev *hdev, char *buf, int len)
> pos += scnprintf(buf + pos, len - pos, "%s", content);
> }
>
> - return 0;
> +out:
> + kfree(data_str);
> + return ret;
> }
>
> static const struct hclge_dbg_item tm_qset_items[] = {

Hi,
could :
pos += scnprintf(buf + pos, len - pos, "%s", <something>);
be more widely used to avoid the alloc()/free() + copy of strings?

CJ

2023-12-05 08:15:49

by Arnd Bergmann

[permalink] [raw]
Subject: Re: [PATCH] [v2] net: hns3: reduce stack usage in hclge_dbg_dump_tm_pri()

On Mon, Dec 4, 2023, at 19:54, Christophe JAILLET wrote:
> Le 04/12/2023 à 09:57, Arnd Bergmann a écrit :
>>
>> Use dynamic allocation for the largest stack object instead. It
>> would be nice to rewrite this file to completely avoid the extra
>> buffer and just use the one that was already allocated by debugfs,
>> but that is a much larger change.
>>
> could :
> pos += scnprintf(buf + pos, len - pos, "%s", <something>);
> be more widely used to avoid the alloc()/free() + copy of strings?

Yes, I think that would help, but this is beyond what I trust
myself to do blindly.

Arnd

2023-12-06 03:23:59

by Jakub Kicinski

[permalink] [raw]
Subject: Re: [PATCH] [v2] net: hns3: reduce stack usage in hclge_dbg_dump_tm_pri()

On Mon, 4 Dec 2023 22:50:55 +0800 Jijie Shao wrote:
> > static int hclge_dbg_dump_tm_pri(struct hclge_dev *hdev, char *buf, int len)
> > {
> > - char data_str[ARRAY_SIZE(tm_pri_items)][HCLGE_DBG_DATA_STR_LEN];
> > + char *data_str;
>
> We want to define variables in an inverted triangle based on the code length.
> so, "char *data_str" should move four lines down.
>
> struct hclge_tm_shaper_para c_shaper_para, p_shaper_para;
> char *result[ARRAY_SIZE(tm_pri_items)], *sch_mode_str;
> char content[HCLGE_DBG_TM_INFO_LEN];
> u8 pri_num, sch_mode, weight, i, j;
> char *data_str;
> int pos, ret;

I took the liberty of fixing this when applying.
Don't want this to fall thru the cracks.

Applied to net-next now, thanks!