kmalloc returns twice the same memory address?

kmalloc returns twice the same memory address?

[Korkakakis Nikos]

Hi there all,

first of all I am writing a kernel module to handle some device using;
uname -a: Linux localhost 2.6.30-gentoo-r1 #1 SMP Sun Jun 28 02:52:23
EEST 2009 i686 AMD Phenom(tm) II X4 940 Processor AuthenticAMD GNU/Linux.
During  the init phase of the module I allocate some memory via
kmalloc() in the following manner;

    omm_dev_buffer = kmalloc(PAGE_SIZE, GFP_KERNEL | GFP_DMA | GFP_ATOMIC);
    if (!omm_dev_buffer) {
        return -ENOMEM;
    }
    clear_buffer (omm_dev_buffer, PAGE_SIZE);
    printk("omm_dev_init: kmalloc'ed omm_dev_buffer OK\n");
    omm_messages = kmalloc(PAGE_SIZE, GFP_KERNEL | GFP_DMA);
    printk ("dev_buff @ 0x%x messages @
0x%x\n",omm_dev_buffer,omm_messages);
    if (!omm_messages) {
        kfree(omm_dev_buffer); /* failing at this point means that the
other allocation succeeded */
        return -ENOMEM;
    }
    clear_buffer (omm_messages, PAGE_SIZE);


where clear_buffer is just a wrapper for memset. My problem is that the
two consecutive memory allocation operations return the same memory
space (both pointers point the same address). If I change the malloc'ed
size ie PAGE_SIZE*2 then the process works as expected. kzalloc behaves
the same way, while alloc_page with the same flags(i.e.
alloc_page(GFP_KERNEL | GFP_DMA | GFP_ATOMIC) ) return a different
memory space and works as expected.

What am I missing here?

Thank's in advance :D

Nikos

Re: kmalloc returns twice the same memory address?

[Robert Hancock]

On 06/30/2009 12:08 PM, Korkakakis Nikos wrote:
> Hi there all,
>
> first of all I am writing a kernel module to handle some device using;
> uname -a: Linux localhost 2.6.30-gentoo-r1 #1 SMP Sun Jun 28 02:52:23
> EEST 2009 i686 AMD Phenom(tm) II X4 940 Processor AuthenticAMD GNU/Linux.
> During  the init phase of the module I allocate some memory via
> kmalloc() in the following manner;
>
>      omm_dev_buffer = kmalloc(PAGE_SIZE, GFP_KERNEL | GFP_DMA | GFP_ATOMIC);

GFP_KERNEL and GFP_ATOMIC are contradictory flags. Also, drivers should 
generally not be using GFP_DMA, if the DMA mapping API is used this is 
not needed.

I don't know if this is what is causing your problem or not.

>      if (!omm_dev_buffer) {
>          return -ENOMEM;
>      }
>      clear_buffer (omm_dev_buffer, PAGE_SIZE);
>      printk("omm_dev_init: kmalloc'ed omm_dev_buffer OK\n");
>      omm_messages = kmalloc(PAGE_SIZE, GFP_KERNEL | GFP_DMA);
>      printk ("dev_buff @ 0x%x messages @
> 0x%x\n",omm_dev_buffer,omm_messages);
>      if (!omm_messages) {
>          kfree(omm_dev_buffer); /* failing at this point means that the
> other allocation succeeded */
>          return -ENOMEM;
>      }
>      clear_buffer (omm_messages, PAGE_SIZE);
>
>
> where clear_buffer is just a wrapper for memset. My problem is that the
> two consecutive memory allocation operations return the same memory
> space (both pointers point the same address). If I change the malloc'ed
> size ie PAGE_SIZE*2 then the process works as expected. kzalloc behaves
> the same way, while alloc_page with the same flags(i.e.
> alloc_page(GFP_KERNEL | GFP_DMA | GFP_ATOMIC) ) return a different
> memory space and works as expected.
>
> What am I missing here?
>
> Thank's in advance :D
>
> Nikos
> --
> To unsubscribe from this list: send the line "unsubscribe linux-newbie" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.linux-learn.org/faqs
>