mtd

mtd

[Amit.Lubovsky]

Hi,
I did as it is said in document
mtd-jffs-HOWTO.txt, v 1.13 2001/05/01 to add MTD support in linux kernel 2.4 for my mips 5kc on my malta board.
I have 2 pcs. flash chips on the board (Intel DT28F160) which has 4 MB capacity.
So I configured xconfig with CFI support, i.e:
CONFIG_MTD
CONFIG_MTD_ROM
CONFIG_MTD_CFI
CONFIG_MTD_CFI_ADVANCED_OPTIONS
CONFIG_MTD_CFI_NOSWAP
CONFIG_MTD_CFI_GEOMETRY
CONFIG_MTD_CFI_B4
CONFIG_MTD_CFI_I4
CONFIG_MTD_CFI_INTELEXT
CONFIG_MTD_CFI_PHYSMAP
CONFIG_MTD_CFI_PHYSMAP_START=1E000000
CONFIG_MTD_CFI_LEN=390900
CONFIG_MTD_CFI_BUSWIDTH=4

since it didn't worked I have change the function ioremap() in physmap.c to ioremap_nocache() but it still doesn't work...
Any clue???

Thanks, Amit.

Re: mtd

[David Woodhouse]

Turn off CONFIG_MTD_CFI_ADVANCED_OPTIONS.

--
dwmw2

mtd

[Levente]

Dear all,


I don't know if I post this to the right place, so forgive my ignorance.

My college discovered a possible bug in OpenWRT's mtd utility. When
writing to jffs device, the 'mtd' utility does not skip bad blocks.

This patch seems to fix this issue.


diff --git a/package/system/mtd/src/jffs2.c b/package/system/mtd/src/jffs2.c
index b432f64ac0..5bf3eec328 100644
--- a/package/system/mtd/src/jffs2.c
+++ b/package/system/mtd/src/jffs2.c
@@ -308,6 +308,16 @@ int mtd_write_jffs2(const char *mtd, const char
*filename, const char *dir)
        for(;;) {
                struct jffs2_unknown_node *node = (struct
jffs2_unknown_node *) buf;

+               while (mtd_block_is_bad(outfd, mtdofs) && (mtdofs < mtdsize)) {
+                       if (!quiet)
+                               fprintf(stderr, "\nSkipping bad block
at 0x%08x   ", mtdofs);
+
+                       mtdofs += erasesize;
+
+                       /* Move the file pointer along over the bad block. */
+                       lseek(outfd, erasesize, SEEK_CUR);
+               }
+
                if (read(outfd, buf, erasesize) != erasesize) {
                        fdeof = 1;
                        break;


I've cloned your utility repository, but didn't find jffs2.c anywhere.

Best regards,
Levente

Re: mtd

[David Oberhollenzer]

Hi,


I took a look at the OpenWRT source repository and the tool you mention looks more
like some home grown utility to me.

The mtd-utils you've cloned are built by package/utils/mtd-utils/Makefile

You may want to sort this out on the OpenWRT mailing list, unless of course the source
of that utility has been copied from mkfs.jffs2 at some point and the later happens
to have the same issue.

See: https://lists.openwrt.org/mailman/listinfo/openwrt-devel


Thanks,

David

Re: mtd

[Levente]

I did this, and the OpenWRT guys directed me here.

But thank you for your help anyways.

Levente

On Fri, May 18, 2018 at 11:09 AM, David Oberhollenzer
<goliath@sigma-star.at> wrote:
> Hi,
>
>
> I took a look at the OpenWRT source repository and the tool you mention looks more
> like some home grown utility to me.
>
> The mtd-utils you've cloned are built by package/utils/mtd-utils/Makefile
>
> You may want to sort this out on the OpenWRT mailing list, unless of course the source
> of that utility has been copied from mkfs.jffs2 at some point and the later happens
> to have the same issue.
>
> See: https://lists.openwrt.org/mailman/listinfo/openwrt-devel
>
>
> Thanks,
>
> David

MTD

[Aneesh]

Hello,
        I am Using an arm9 processor(kernel 2.6.24) and tried to
implement JFFS2 in parallel flash for data storage.but it failed .
my parallel flash is AT49BV642DT .First i erased the flash and mounted
to a drive .then i created a file there .After that i unmounted the
drive .and
 when i am trying to mount the jffs2 again  ,It is showing an error like
follows
"Magic bitmask 0x1985 not found at 0x007e0078:  0x6d63 instead  "
can you help me to solve this problem ?
                                                    Regards
                                                            Aneesh.

Re: MTD

[Andrey Yurovsky]

Please see the FAQ:
http://www.linux-mtd.infradead.org/faq/jffs2.html#L_magicnfound

On Tue, Mar 25, 2008 at 8:58 PM, Aneesh <aneesh_g@cms.com> wrote:
>
>
>  Hello,
>         I am Using an arm9 processor(kernel 2.6.24) and tried to
>  implement JFFS2 in parallel flash for data storage.but it failed .
>  my parallel flash is AT49BV642DT .First i erased the flash and mounted
>  to a drive .then i created a file there .After that i unmounted the
>  drive .and
>   when i am trying to mount the jffs2 again  ,It is showing an error like
>  follows
>  "Magic bitmask 0x1985 not found at 0x007e0078:  0x6d63 instead  "
>  can you help me to solve this problem ?
>                                                     Regards
>                                                             Aneesh.
>
>
>  ______________________________________________________
>  Linux MTD discussion mailing list
>  http://lists.infradead.org/mailman/listinfo/linux-mtd/
>



-- 
Andrey Yurovsky
cozybit Inc.
p 415 974 6770 x14
f 415 974 6771
e andrey@cozybit.com

Mtd

[yogesh shivabasappa]

Hi 

I'm new to mtd devices. I'm using intelp30 Nor flash memory chip
in one of my project . i am accessing this as a character device
I wanted to know how this deffers from the normal character
devices

thanks 

yogesh

________________________________________________
Get your own "800" number
Voicemail, fax, email, and a lot more
http://www.ureach.com/reg/tag

mtd

[Bob Canup]

 I would like to possibly add something to the "to do" list:

Make sure the code works with older kernels. Much development work on
embedded systems is done on the proven stable 2.0 and 2.2 kernels.
Embedded systems tend not to have bleeding edge software in them for
reliability reasons. That is part of the reason that Alan Cox still
turns out 2.0.x kernels.

Bob




To unsubscribe, send "unsubscribe mtd" to majordomo@infradead.org

Re: mtd

[David Woodhouse]

rcanup@go2fax.com said:
>  Make sure the code works with older kernels. Much development work on
> embedded systems is done on the proven stable 2.0 and 2.2 kernels.
> Embedded systems tend not to have bleeding edge software in them for
> reliability reasons. That is part of the reason that Alan Cox still
> turns out 2.0.x kernels.

Agreed. Making it work with 2.2 is sort of implicit - I know Trevor's 
developing on 2.2, and I try to test with 2.2 periodically. Back-porting to 
2.0 could possibly be considered a separate task.

--
dwmw2




To unsubscribe, send "unsubscribe mtd" to majordomo@infradead.org