Hi, Arun and Pramod, Yes I knew this is from the page fault handling routine. (In my case, unhandled_fault function insdie do_sparc_fault function) I know before the page fault, the pmd value (the value pointed to by pgd of context 1) was zero, so page fault occured. This is the message I side-printed on the UART terminal. (If the frame buffer access was successful, the printk would have been on LCD but it didn't print on the LCD because of this paging error.) The page fault handler should make a new pte for the frame buffer address but giving me this error. Unable to handel kernel paging request at virtual address 696e1000 {mm,active_mm}->context = 00000001 {mm,active_mm}->pgd = c800a400 In what case can this happen? Any clue will be very helpful. Thanks! Chan ________________________________ From : "Arun KS" <getarunks@gmail.com> Sent : 2014-03-27 18:27:57 ( +09:00 ) To : Chan Kim <ckim@etri.re.kr> Cc : kernelnewbies@kernelnewbies.org <kernelnewbies@kernelnewbies.org> Subject : Re: Unable to handle kernel paging request at virtual address ... while writing to frame buffer.. Hi Chan, On Thu, Mar 27, 2014 at 2:25 PM, Chan Kim wrote:

Hi, I have this 'unable to handle kernel paging request at virtual address ...' This means the virtual address used doesn't have a proper mapping in the page table.

Thanks, Arun

when writing to the frame buffer. Has anybody seen similar situation? and hopeful can tell me what the problem was? Thanks! Chan

_______________________________________________ Kernelnewbies mailing list Kernelnewbies@kernelnewbies.org http://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies

During the initialization, the kernel writes to the frame buffer using printk and I can see it on the LCD. But when I am writing in the busybox (which is user program seen from the kernel) using 'printf', this page error occurs. For both printk and printf cases, the context table base is the same, but one(printk) is in context 0 while the otehr is in context 1. (but all the contexts have the same pgd values in both cases, each with different value for printk and printf because they are different process. and different threads in a process have different context I understand). The frame buffer address appears the same in both printk and printf. I've got the frame buffer address using dma_alloc_writecombine function (I tried with dma_alloc_coherent, but it fails with error in our system). I guess there should be some typical cases when this happens. (I know the page fault routine should assign a virtual address for the page frame containing the frame buffer and connect it to the page table for the context but it could not handle it. I will delve into it anyway..) Any hint? I hope somebody could shed some light to this problem. Thanks, Chan ________________________________ From : "Valdis.Kletnieks@vt.edu" <Valdis.Kletnieks@vt.edu> Sent : 2014-03-28 01:29:21 ( +09:00 ) To : Chan Kim <ckim@etri.re.kr> Cc : kernelnewbies@kernelnewbies.org <kernelnewbies@kernelnewbies.org> Subject : Re: Unable to handle kernel paging request at virtual address ... while writing to frame buffer.. On Thu, 27 Mar 2014 08:55:10 -0000, Chan Kim said:

I have this 'unable to handle kernel paging request at virtual address ...' when writing to the frame buffer.

That almost certainly means you tried to write to something other than the actual frame buffer.