[PATCH 2/2] virtio-console: remove unnecessary kmemdup()

[PATCH 2/2] virtio-console: remove unnecessary kmemdup()

[Xianting Tian]

hvc framework will never pass stack memory to the put_chars() function,
So the calling of kmemdup() is unnecessary, remove it.

Fixes: c4baad5029 ("virtio-console: avoid DMA from stack")
Signed-off-by: Xianting Tian <xianting.tian@linux.alibaba.com>
---
 drivers/char/virtio_console.c | 12 ++----------
 1 file changed, 2 insertions(+), 10 deletions(-)

diff --git a/drivers/char/virtio_console.c b/drivers/char/virtio_console.c
index 7eaf303a7..4ed3ffb1d 100644
--- a/drivers/char/virtio_console.c
+++ b/drivers/char/virtio_console.c
@@ -1117,8 +1117,6 @@ static int put_chars(u32 vtermno, const char *buf, int count)
 {
 	struct port *port;
 	struct scatterlist sg[1];
-	void *data;
-	int ret;
 
 	if (unlikely(early_put_chars))
 		return early_put_chars(vtermno, buf, count);
@@ -1127,14 +1125,8 @@ static int put_chars(u32 vtermno, const char *buf, int count)
 	if (!port)
 		return -EPIPE;
 
-	data = kmemdup(buf, count, GFP_ATOMIC);
-	if (!data)
-		return -ENOMEM;
-
-	sg_init_one(sg, data, count);
-	ret = __send_to_port(port, sg, 1, count, data, false);
-	kfree(data);
-	return ret;
+	sg_init_one(sg, buf, count);
+	return __send_to_port(port, sg, 1, count, (void *)buf, false);
 }
 
 /*
-- 
2.17.1

[PATCH 2/2] virtio-console: remove unnecessary kmemdup()

[Xianting Tian]

hvc framework will never pass stack memory to the put_chars() function,
So the calling of kmemdup() is unnecessary, remove it.

Fixes: c4baad5029 ("virtio-console: avoid DMA from stack")
Signed-off-by: Xianting Tian <xianting.tian@linux.alibaba.com>
---
 drivers/char/virtio_console.c | 12 ++----------
 1 file changed, 2 insertions(+), 10 deletions(-)

diff --git a/drivers/char/virtio_console.c b/drivers/char/virtio_console.c
index 7eaf303a7..4ed3ffb1d 100644
--- a/drivers/char/virtio_console.c
+++ b/drivers/char/virtio_console.c
@@ -1117,8 +1117,6 @@ static int put_chars(u32 vtermno, const char *buf, int count)
 {
 	struct port *port;
 	struct scatterlist sg[1];
-	void *data;
-	int ret;
 
 	if (unlikely(early_put_chars))
 		return early_put_chars(vtermno, buf, count);
@@ -1127,14 +1125,8 @@ static int put_chars(u32 vtermno, const char *buf, int count)
 	if (!port)
 		return -EPIPE;
 
-	data = kmemdup(buf, count, GFP_ATOMIC);
-	if (!data)
-		return -ENOMEM;
-
-	sg_init_one(sg, data, count);
-	ret = __send_to_port(port, sg, 1, count, data, false);
-	kfree(data);
-	return ret;
+	sg_init_one(sg, buf, count);
+	return __send_to_port(port, sg, 1, count, (void *)buf, false);
 }
 
 /*
-- 
2.17.1

Re: [PATCH 2/2] virtio-console: remove unnecessary kmemdup()

[Jiri Slaby]

Hi,

why is this 2/2? I seem (Lore neither) to find 1/2.

On 01. 08. 21, 7:16, Xianting Tian wrote:
> hvc framework will never pass stack memory to the put_chars() function,

Am I blind or missing something?

hvc_console_print(...)
{
   char c[N_OUTBUF]
...
   cons_ops[index]->put_chars(vtermnos[index], c, i);

The same here:

hvc_poll_put_char(..., char ch)
{
...
    n = hp->ops->put_chars(hp->vtermno, &ch, 1);

AFAICS both of them *pass* a pointer to stack variable.

> So the calling of kmemdup() is unnecessary, remove it.
> 
> Fixes: c4baad5029 ("virtio-console: avoid DMA from stack")

This patch doesn't "Fix" -- it reverts the commit. You should've CCed 
the author too.

> Signed-off-by: Xianting Tian <xianting.tian@linux.alibaba.com>
> ---
>   drivers/char/virtio_console.c | 12 ++----------
>   1 file changed, 2 insertions(+), 10 deletions(-)
> 
> diff --git a/drivers/char/virtio_console.c b/drivers/char/virtio_console.c
> index 7eaf303a7..4ed3ffb1d 100644
> --- a/drivers/char/virtio_console.c
> +++ b/drivers/char/virtio_console.c
> @@ -1117,8 +1117,6 @@ static int put_chars(u32 vtermno, const char *buf, int count)
>   {
>   	struct port *port;
>   	struct scatterlist sg[1];
> -	void *data;
> -	int ret;
>   
>   	if (unlikely(early_put_chars))
>   		return early_put_chars(vtermno, buf, count);
> @@ -1127,14 +1125,8 @@ static int put_chars(u32 vtermno, const char *buf, int count)
>   	if (!port)
>   		return -EPIPE;
>   
> -	data = kmemdup(buf, count, GFP_ATOMIC);
> -	if (!data)
> -		return -ENOMEM;
> -
> -	sg_init_one(sg, data, count);
> -	ret = __send_to_port(port, sg, 1, count, data, false);
> -	kfree(data);
> -	return ret;
> +	sg_init_one(sg, buf, count);
> +	return __send_to_port(port, sg, 1, count, (void *)buf, false);
>   }
>   
>   /*
> 


-- 
js
suse labs

Re: [PATCH 2/2] virtio-console: remove unnecessary kmemdup()

[Jiri Slaby]

Hi,

why is this 2/2? I seem (Lore neither) to find 1/2.

On 01. 08. 21, 7:16, Xianting Tian wrote:
> hvc framework will never pass stack memory to the put_chars() function,

Am I blind or missing something?

hvc_console_print(...)
{
   char c[N_OUTBUF]
...
   cons_ops[index]->put_chars(vtermnos[index], c, i);

The same here:

hvc_poll_put_char(..., char ch)
{
...
    n = hp->ops->put_chars(hp->vtermno, &ch, 1);

AFAICS both of them *pass* a pointer to stack variable.

> So the calling of kmemdup() is unnecessary, remove it.
> 
> Fixes: c4baad5029 ("virtio-console: avoid DMA from stack")

This patch doesn't "Fix" -- it reverts the commit. You should've CCed 
the author too.

> Signed-off-by: Xianting Tian <xianting.tian@linux.alibaba.com>
> ---
>   drivers/char/virtio_console.c | 12 ++----------
>   1 file changed, 2 insertions(+), 10 deletions(-)
> 
> diff --git a/drivers/char/virtio_console.c b/drivers/char/virtio_console.c
> index 7eaf303a7..4ed3ffb1d 100644
> --- a/drivers/char/virtio_console.c
> +++ b/drivers/char/virtio_console.c
> @@ -1117,8 +1117,6 @@ static int put_chars(u32 vtermno, const char *buf, int count)
>   {
>   	struct port *port;
>   	struct scatterlist sg[1];
> -	void *data;
> -	int ret;
>   
>   	if (unlikely(early_put_chars))
>   		return early_put_chars(vtermno, buf, count);
> @@ -1127,14 +1125,8 @@ static int put_chars(u32 vtermno, const char *buf, int count)
>   	if (!port)
>   		return -EPIPE;
>   
> -	data = kmemdup(buf, count, GFP_ATOMIC);
> -	if (!data)
> -		return -ENOMEM;
> -
> -	sg_init_one(sg, data, count);
> -	ret = __send_to_port(port, sg, 1, count, data, false);
> -	kfree(data);
> -	return ret;
> +	sg_init_one(sg, buf, count);
> +	return __send_to_port(port, sg, 1, count, (void *)buf, false);
>   }
>   
>   /*
> 


-- 
js
suse labs

Re: [PATCH 2/2] virtio-console: remove unnecessary kmemdup()

[Xianting Tian]

[-- Attachment #1: Type: text/plain, Size: 2454 bytes --]


在 2021/8/2 下午3:25, Jiri Slaby 写道:
> Hi,
>
> why is this 2/2? I seem (Lore neither) to find 1/2.
You didn't receive 1/2?
[PATCH 1/2] tty: hvc: pass DMA capable memory to put_chars()
https://lkml.org/lkml/2021/8/1/8 <https://lkml.org/lkml/2021/8/1/8>
>
> On 01. 08. 21, 7:16, Xianting Tian wrote:
>> hvc framework will never pass stack memory to the put_chars() function,
>
> Am I blind or missing something?
>
> hvc_console_print(...)
> {
>   char c[N_OUTBUF]
> ...
>   cons_ops[index]->put_chars(vtermnos[index], c, i);
>
> The same here:
>
> hvc_poll_put_char(..., char ch)
> {
> ...
>    n = hp->ops->put_chars(hp->vtermno, &ch, 1);
>
> AFAICS both of them *pass* a pointer to stack variable.

yes, I discussed the issue with Arnd before in below thread,  you can 
get the history, thanks

https://lkml.org/lkml/2021/7/27/494 <https://lkml.org/lkml/2021/7/27/494>

>
>> So the calling of kmemdup() is unnecessary, remove it.
>>
>> Fixes: c4baad5029 ("virtio-console: avoid DMA from stack")
>
> This patch doesn't "Fix" -- it reverts the commit. You should've CCed 
> the author too.

yes, we discussed ther issue in above thread, which we CCed the author.

>
>> Signed-off-by: Xianting Tian <xianting.tian@linux.alibaba.com>
>> ---
>>   drivers/char/virtio_console.c | 12 ++----------
>>   1 file changed, 2 insertions(+), 10 deletions(-)
>>
>> diff --git a/drivers/char/virtio_console.c 
>> b/drivers/char/virtio_console.c
>> index 7eaf303a7..4ed3ffb1d 100644
>> --- a/drivers/char/virtio_console.c
>> +++ b/drivers/char/virtio_console.c
>> @@ -1117,8 +1117,6 @@ static int put_chars(u32 vtermno, const char 
>> *buf, int count)
>>   {
>>       struct port *port;
>>       struct scatterlist sg[1];
>> -    void *data;
>> -    int ret;
>>         if (unlikely(early_put_chars))
>>           return early_put_chars(vtermno, buf, count);
>> @@ -1127,14 +1125,8 @@ static int put_chars(u32 vtermno, const char 
>> *buf, int count)
>>       if (!port)
>>           return -EPIPE;
>>   -    data = kmemdup(buf, count, GFP_ATOMIC);
>> -    if (!data)
>> -        return -ENOMEM;
>> -
>> -    sg_init_one(sg, data, count);
>> -    ret = __send_to_port(port, sg, 1, count, data, false);
>> -    kfree(data);
>> -    return ret;
>> +    sg_init_one(sg, buf, count);
>> +    return __send_to_port(port, sg, 1, count, (void *)buf, false);
>>   }
>>     /*
>>
>
>

[-- Attachment #2: Type: text/html, Size: 4975 bytes --]

Re: [PATCH 2/2] virtio-console: remove unnecessary kmemdup()

[Jiri Slaby]

On 02. 08. 21, 10:32, Xianting Tian wrote:
> 
> 在 2021/8/2 下午3:25, Jiri Slaby 写道:
>> Hi,
>>
>> why is this 2/2? I seem (Lore neither) to find 1/2.
> You didn't receive 1/2?
> [PATCH 1/2] tty: hvc: pass DMA capable memory to put_chars()
> https://lkml.org/lkml/2021/8/1/8 <https://lkml.org/lkml/2021/8/1/8>

Oh, I did, but it's not properly threaded. PLease fix your setup.

>> On 01. 08. 21, 7:16, Xianting Tian wrote:
>>> hvc framework will never pass stack memory to the put_chars() function,
>>
>> Am I blind or missing something?
>>
>> hvc_console_print(...)
>> {
>>   char c[N_OUTBUF]
>> ...
>>   cons_ops[index]->put_chars(vtermnos[index], c, i);
>>
>> The same here:
>>
>> hvc_poll_put_char(..., char ch)
>> {
>> ...
>>    n = hp->ops->put_chars(hp->vtermno, &ch, 1);
>>
>> AFAICS both of them *pass* a pointer to stack variable.
> 
> yes, I discussed the issue with Arnd before in below thread,  you can 
> get the history, thanks
> 
> https://lkml.org/lkml/2021/7/27/494 <https://lkml.org/lkml/2021/7/27/494>

So is this a v2? You should have noted that. And what changed from v1 too.

>>> So the calling of kmemdup() is unnecessary, remove it.
>>>
>>> Fixes: c4baad5029 ("virtio-console: avoid DMA from stack")
>>
>> This patch doesn't "Fix" -- it reverts the commit. You should've CCed 
>> the author too.
> 
> yes, we discussed ther issue in above thread, which we CCed the author.

I don't see any input from the author?


Anyway, 1/2 does not even build, so you will send v3 with all the above 
fixed, hopefully.

thanks,
-- 
js

Re: [PATCH 2/2] virtio-console: remove unnecessary kmemdup()

[Jiri Slaby]

On 02. 08. 21, 10:32, Xianting Tian wrote:
> 
> 在 2021/8/2 下午3:25, Jiri Slaby 写道:
>> Hi,
>>
>> why is this 2/2? I seem (Lore neither) to find 1/2.
> You didn't receive 1/2?
> [PATCH 1/2] tty: hvc: pass DMA capable memory to put_chars()
> https://lkml.org/lkml/2021/8/1/8 <https://lkml.org/lkml/2021/8/1/8>

Oh, I did, but it's not properly threaded. PLease fix your setup.

>> On 01. 08. 21, 7:16, Xianting Tian wrote:
>>> hvc framework will never pass stack memory to the put_chars() function,
>>
>> Am I blind or missing something?
>>
>> hvc_console_print(...)
>> {
>>   char c[N_OUTBUF]
>> ...
>>   cons_ops[index]->put_chars(vtermnos[index], c, i);
>>
>> The same here:
>>
>> hvc_poll_put_char(..., char ch)
>> {
>> ...
>>    n = hp->ops->put_chars(hp->vtermno, &ch, 1);
>>
>> AFAICS both of them *pass* a pointer to stack variable.
> 
> yes, I discussed the issue with Arnd before in below thread,  you can 
> get the history, thanks
> 
> https://lkml.org/lkml/2021/7/27/494 <https://lkml.org/lkml/2021/7/27/494>

So is this a v2? You should have noted that. And what changed from v1 too.

>>> So the calling of kmemdup() is unnecessary, remove it.
>>>
>>> Fixes: c4baad5029 ("virtio-console: avoid DMA from stack")
>>
>> This patch doesn't "Fix" -- it reverts the commit. You should've CCed 
>> the author too.
> 
> yes, we discussed ther issue in above thread, which we CCed the author.

I don't see any input from the author?


Anyway, 1/2 does not even build, so you will send v3 with all the above 
fixed, hopefully.

thanks,
-- 
js

Re: [PATCH 2/2] virtio-console: remove unnecessary kmemdup()

[Xianting Tian]

在 2021/8/2 下午4:40, Jiri Slaby 写道:
> On 02. 08. 21, 10:32, Xianting Tian wrote:
>>
>> 在 2021/8/2 下午3:25, Jiri Slaby 写道:
>>> Hi,
>>>
>>> why is this 2/2? I seem (Lore neither) to find 1/2.
>> You didn't receive 1/2?
>> [PATCH 1/2] tty: hvc: pass DMA capable memory to put_chars()
>> https://lkml.org/lkml/2021/8/1/8 <https://lkml.org/lkml/2021/8/1/8>
>
> Oh, I did, but it's not properly threaded. PLease fix your setup.
Ok, thanks
>
>>> On 01. 08. 21, 7:16, Xianting Tian wrote:
>>>> hvc framework will never pass stack memory to the put_chars() 
>>>> function,
>>>
>>> Am I blind or missing something?
>>>
>>> hvc_console_print(...)
>>> {
>>>   char c[N_OUTBUF]
>>> ...
>>>   cons_ops[index]->put_chars(vtermnos[index], c, i);
>>>
>>> The same here:
>>>
>>> hvc_poll_put_char(..., char ch)
>>> {
>>> ...
>>>    n = hp->ops->put_chars(hp->vtermno, &ch, 1);
>>>
>>> AFAICS both of them *pass* a pointer to stack variable.
>>
>> yes, I discussed the issue with Arnd before in below thread, you can 
>> get the history, thanks
>>
>> https://lkml.org/lkml/2021/7/27/494 
>> <https://lkml.org/lkml/2021/7/27/494>
>
> So is this a v2? You should have noted that. And what changed from v1 
> too.

I think yes, I should mentioned it in this patch, sorry for that:(

>
>>>> So the calling of kmemdup() is unnecessary, remove it.
>>>>
>>>> Fixes: c4baad5029 ("virtio-console: avoid DMA from stack")
>>>
>>> This patch doesn't "Fix" -- it reverts the commit. You should've 
>>> CCed the author too.
>>
>> yes, we discussed ther issue in above thread, which we CCed the author.
>
> I don't see any input from the author?
>
>
> Anyway, 1/2 does not even build, so you will send v3 with all the 
> above fixed, hopefully.

yes, I will send v3 patch after I figured out a better solution based on 
Arnd's comments for the patch '1/2'.

Do you have any other suggestion for the solution?

thanks.

>
> thanks,

Re: [PATCH 2/2] virtio-console: remove unnecessary kmemdup()

[Xianting Tian]

在 2021/8/2 下午4:40, Jiri Slaby 写道:
> On 02. 08. 21, 10:32, Xianting Tian wrote:
>>
>> 在 2021/8/2 下午3:25, Jiri Slaby 写道:
>>> Hi,
>>>
>>> why is this 2/2? I seem (Lore neither) to find 1/2.
>> You didn't receive 1/2?
>> [PATCH 1/2] tty: hvc: pass DMA capable memory to put_chars()
>> https://lkml.org/lkml/2021/8/1/8 <https://lkml.org/lkml/2021/8/1/8>
>
> Oh, I did, but it's not properly threaded. PLease fix your setup.
Ok, thanks
>
>>> On 01. 08. 21, 7:16, Xianting Tian wrote:
>>>> hvc framework will never pass stack memory to the put_chars() 
>>>> function,
>>>
>>> Am I blind or missing something?
>>>
>>> hvc_console_print(...)
>>> {
>>>   char c[N_OUTBUF]
>>> ...
>>>   cons_ops[index]->put_chars(vtermnos[index], c, i);
>>>
>>> The same here:
>>>
>>> hvc_poll_put_char(..., char ch)
>>> {
>>> ...
>>>    n = hp->ops->put_chars(hp->vtermno, &ch, 1);
>>>
>>> AFAICS both of them *pass* a pointer to stack variable.
>>
>> yes, I discussed the issue with Arnd before in below thread, you can 
>> get the history, thanks
>>
>> https://lkml.org/lkml/2021/7/27/494 
>> <https://lkml.org/lkml/2021/7/27/494>
>
> So is this a v2? You should have noted that. And what changed from v1 
> too.

I think yes, I should mentioned it in this patch, sorry for that:(

>
>>>> So the calling of kmemdup() is unnecessary, remove it.
>>>>
>>>> Fixes: c4baad5029 ("virtio-console: avoid DMA from stack")
>>>
>>> This patch doesn't "Fix" -- it reverts the commit. You should've 
>>> CCed the author too.
>>
>> yes, we discussed ther issue in above thread, which we CCed the author.
>
> I don't see any input from the author?
>
>
> Anyway, 1/2 does not even build, so you will send v3 with all the 
> above fixed, hopefully.

yes, I will send v3 patch after I figured out a better solution based on 
Arnd's comments for the patch '1/2'.

Do you have any other suggestion for the solution?

thanks.

>
> thanks,