Loading ...
Sorry, an error occurred while loading the content.

Dumb-ass Yaffs from 2.6 incompatible with 2.4

Expand Messages
  • j.chitte
    new topic ??
    Message 1 of 6 , May 29, 2008
    View Source
    • 0 Attachment
      new topic ??
    • j.chitte
      ... A finally conviced this dumb-assed group software to create a new topic when I click Start topic . now where were we ?? ... Well, it isn t yet ... ... the
      Message 2 of 6 , May 29, 2008
      View Source
      • 0 Attachment
        --- In ts-7000@yahoogroups.com, "j.chitte" <j.chitte@...> wrote:
        >
        > new topic ??
        >

        A finally conviced this dumb-assed group software to create a new
        topic when I click "Start topic".

        now where were we ??



        --- In ts-7000@yahoogroups.com, "bsmuck2000" <bsmucker@...> wrote:
        >
        > (Hopefully this is a new thread this time)

        Well, it isn't yet ...

        >
        > Hello,
        >
        > We are working at transitioning our ts7260-based system to using
        the 2.6
        > kernel (right now the one supplied by Technologics. One problem that
        > I'm running into is that when I boot to 2.6, the yaffs2 file system
        on

        Was the ts-7260 delivered with YAFFS or YAFFS2? IIRC, YAFFS2 was used
        only for the larger flash sizes (2k page sizes).

        > the flash is not visible. I can mount it fine, but nothing is there.

        Anything in the kernel log after the mount or attempted read?

        > If I rewrite the filesystem while in 2.6, it works fine, but when I
        boot
        > back into 2.4, it again shows a blank filesystem.
        >
        > Is anyone aware of this incompatibility and how to fix it?

        Check the YAFFS2 build options on the 2.4 config, and make sure your
        2.6 version is configured the same or at least backwards compatible.

        Especially look at YAFFS_AUTO_YAFFS2, YAFFS_YAFFS2,
        YAFFS_DISABLE_WIDE_TNODES, and their help text in Kconfig. If you are
        mixing yaffs and yaffs2, these config options can help make this
        transparent.

        You can also turn on some debugging options in yaffs_fs.c (need to
        recompile).

        >
        > For the transition time, it is essential that we be able to access
        the
        > yaffs filesystem from both 2.4 and 2.6.
        >
        > Thanks,
        >
        > Brian
        >

        Regards, .......... Charlie




        Thu May 29, 2008 8:52 pm

        Show Message Info
        "charliem_1216" <charliem_1216@...>
        charliem_1216
        Offline
        Send Email

        Forward < Prev Message | Next Message >
      • adam_kaczynski
        Hi All, There is a potential issue on 7260 which may result in wiping out RedBoot if 2.6 is used on the board with default settings. The 2.6 Linux root
        Message 3 of 6 , Jun 2, 2008
        View Source
        • 0 Attachment
          Hi All,

          There is a potential issue on 7260 which may result in wiping out
          RedBoot if 2.6 is used on the board with default settings.

          The 2.6 Linux root partition overlaps with the RedBoot partition
          defined for 2.4

          This is what is printed out when 2.4 is starting:

          Using static partition definition

          Creating 3 MTD partitions on "NAND 32MiB 3,3V 8-bit":

          0x00000000-0x00004000 : "TS-BOOTROM"

          0x00004000-0x01d04000 : "Linux"

          0x01d04000-0x02000000 : "RedBoot"

          And this is printed out when 2.6 is starting:

          Using static partition definition

          Creating 3 MTD partitions on "ts7250-nand":

          0x00000000-0x00020000 : "TS-BOOTROM"

          0x00020000-0x01d20000 : "Linux"

          0x01d20000-0x0201c000 : "RedBoot"

          mtd: partition "RedBoot" extends beyond the end of device
          "ts7250-nand" -- size truncated to 0x2e0000

          The memory definition found in
          linux-2.6.whatever/drivers/mtd/nand/ts7250.c is inconsistent with 2.4

          This is the offending structure:

          /*
          * Define static partitions for flash device
          */
          static struct mtd_partition partition_info32[] = {
          {
          .name = "TS-BOOTROM",
          .offset = 0x00000000,
          .size = 0x00020000,
          }, {
          .name = "Linux",
          .offset = 0x00020000,
          .size = 0x01d00000,
          }, {
          .name = "RedBoot",
          .offset = 0x01d20000,
          .size = 0x002fc000,
          },
          };

          After adjusting the settings to reflect the reality (on the flash) the
          error message about truncation has disappeared.

          However:
          There is still something wrong possibly with the yaffs settings and I
          cannot use both versions of the kernel to access the same data on
          Flash. I may not need to do so since I have other options in my
          project but if I make it working I will obviously post an update.

          Regards,
          Adam

          --- In ts-7000@yahoogroups.com, "j.chitte" <j.chitte@...> wrote:
          >
          > --- In ts-7000@yahoogroups.com, "j.chitte" <j.chitte@> wrote:
          > >
          > > new topic ??
          > >
          >
          > A finally conviced this dumb-assed group software to create a new
          > topic when I click "Start topic".
          >
          > now where were we ??
          >
          >
          >
          > --- In ts-7000@yahoogroups.com, "bsmuck2000" <bsmucker@> wrote:
          > >
          > > (Hopefully this is a new thread this time)
          >
          > Well, it isn't yet ...
          >
          > >
          > > Hello,
          > >
          > > We are working at transitioning our ts7260-based system to using
          > the 2.6
          > > kernel (right now the one supplied by Technologics. One problem that
          > > I'm running into is that when I boot to 2.6, the yaffs2 file system
          > on
          >
          > Was the ts-7260 delivered with YAFFS or YAFFS2? IIRC, YAFFS2 was used
          > only for the larger flash sizes (2k page sizes).
          >
          > > the flash is not visible. I can mount it fine, but nothing is there.
          >
          > Anything in the kernel log after the mount or attempted read?
          >
          > > If I rewrite the filesystem while in 2.6, it works fine, but when I
          > boot
          > > back into 2.4, it again shows a blank filesystem.
          > >
          > > Is anyone aware of this incompatibility and how to fix it?
          >
          > Check the YAFFS2 build options on the 2.4 config, and make sure your
          > 2.6 version is configured the same or at least backwards compatible.
          >
          > Especially look at YAFFS_AUTO_YAFFS2, YAFFS_YAFFS2,
          > YAFFS_DISABLE_WIDE_TNODES, and their help text in Kconfig. If you are
          > mixing yaffs and yaffs2, these config options can help make this
          > transparent.
          >
          > You can also turn on some debugging options in yaffs_fs.c (need to
          > recompile).
          >
          > >
          > > For the transition time, it is essential that we be able to access
          > the
          > > yaffs filesystem from both 2.4 and 2.6.
          > >
          > > Thanks,
          > >
          > > Brian
          > >
          >
          > Regards, .......... Charlie
          >
          >
          >
          >
          > Thu May 29, 2008 8:52 pm
          >
          > Show Message Info
          > "charliem_1216" <charliem_1216@>
          > charliem_1216
          > Offline
          > Send Email
          >
          > Forward < Prev Message | Next Message >
          >
        • j.chitte
          ... definition ... BOOTROM ... Linux ... definition ... BOOTROM ... Linux ... RedBoot ... 2.4 ... the ... and I ... new ... using ... problem that ...
          Message 4 of 6 , Jun 2, 2008
          View Source
          • 0 Attachment
            --- In ts-7000@yahoogroups.com,
            "adam_kaczynski" <adam_kaczynski@...> wrote:
            >
            > Hi All,
            >
            > There is a potential issue on 7260 which may result in wiping out
            > RedBoot if 2.6 is used on the board with default settings.
            >
            > The 2.6 Linux root partition overlaps with the RedBoot partition
            > defined for 2.4
            >
            > This is what is printed out when 2.4 is starting:
            >
            > Using static partition
            definition
            >
            > Creating 3 MTD partitions on "NAND 32MiB 3,3V 8-
            bit":
            >
            > 0x00000000-0x00004000 : "TS-
            BOOTROM"
            >
            > 0x00004000-0x01d04000 :
            "Linux"
            >
            > 0x01d04000-0x02000000 : "RedBoot"
            >
            > And this is printed out when 2.6 is starting:
            >
            > Using static partition
            definition
            >
            > Creating 3 MTD partitions on "ts7250-
            nand":
            >
            > 0x00000000-0x00020000 : "TS-
            BOOTROM"
            >
            > 0x00020000-0x01d20000 :
            "Linux"
            >
            > 0x01d20000-0x0201c000 :
            "RedBoot"
            >
            > mtd: partition "RedBoot" extends beyond the end of device
            > "ts7250-nand" -- size truncated to 0x2e0000
            >
            > The memory definition found in
            > linux-2.6.whatever/drivers/mtd/nand/ts7250.c is inconsistent with
            2.4
            >
            > This is the offending structure:
            >
            > /*
            > * Define static partitions for flash device
            > */
            > static struct mtd_partition partition_info32[] = {
            > {
            > .name = "TS-BOOTROM",
            > .offset = 0x00000000,
            > .size = 0x00020000,
            > }, {
            > .name = "Linux",
            > .offset = 0x00020000,
            > .size = 0x01d00000,
            > }, {
            > .name = "RedBoot",
            > .offset = 0x01d20000,
            > .size = 0x002fc000,
            > },
            > };
            >
            > After adjusting the settings to reflect the reality (on the flash)
            the
            > error message about truncation has disappeared.
            >
            > However:
            > There is still something wrong possibly with the yaffs settings
            and I
            > cannot use both versions of the kernel to access the same data on
            > Flash. I may not need to do so since I have other options in my
            > project but if I make it working I will obviously post an update.
            >
            > Regards,
            > Adam
            >
            > --- In ts-7000@yahoogroups.com, "j.chitte" <j.chitte@> wrote:
            > >
            > > --- In ts-7000@yahoogroups.com, "j.chitte" <j.chitte@> wrote:
            > > >
            > > > new topic ??
            > > >
            > >
            > > A finally conviced this dumb-assed group software to create a
            new
            > > topic when I click "Start topic".
            > >
            > > now where were we ??
            > >
            > >
            > >
            > > --- In ts-7000@yahoogroups.com, "bsmuck2000" <bsmucker@> wrote:
            > > >
            > > > (Hopefully this is a new thread this time)
            > >
            > > Well, it isn't yet ...
            > >
            > > >
            > > > Hello,
            > > >
            > > > We are working at transitioning our ts7260-based system to
            using
            > > the 2.6
            > > > kernel (right now the one supplied by Technologics. One
            problem that
            > > > I'm running into is that when I boot to 2.6, the yaffs2 file
            system
            > > on
            > >
            > > Was the ts-7260 delivered with YAFFS or YAFFS2? IIRC, YAFFS2 was
            used
            > > only for the larger flash sizes (2k page sizes).
            > >
            > > > the flash is not visible. I can mount it fine, but nothing is
            there.
            > >
            > > Anything in the kernel log after the mount or attempted read?
            > >
            > > > If I rewrite the filesystem while in 2.6, it works fine, but
            when I
            > > boot
            > > > back into 2.4, it again shows a blank filesystem.
            > > >
            > > > Is anyone aware of this incompatibility and how to fix it?
            > >
            > > Check the YAFFS2 build options on the 2.4 config, and make sure
            your
            > > 2.6 version is configured the same or at least backwards
            compatible.
            > >
            > > Especially look at YAFFS_AUTO_YAFFS2, YAFFS_YAFFS2,
            > > YAFFS_DISABLE_WIDE_TNODES, and their help text in Kconfig. If
            you are
            > > mixing yaffs and yaffs2, these config options can help make this
            > > transparent.
            > >
            > > You can also turn on some debugging options in yaffs_fs.c (need
            to
            > > recompile).
            > >
            > > >
            > > > For the transition time, it is essential that we be able to
            access
            > > the
            > > > yaffs filesystem from both 2.4 and 2.6.
            > > >
            > > > Thanks,
            > > >
            > > > Brian
            > > >
            > >
            > > Regards, .......... Charlie
            > >
            > >
            > >
            > >
            > > Thu May 29, 2008 8:52 pm
            > >
            > > Show Message Info
            > > "charliem_1216" <charliem_1216@>
            > > charliem_1216
            > > Offline
            > > Send Email
            > >
            > > Forward < Prev Message | Next Message >
            > >
            >

            Hi ,

            Triffid hunter posted a patch for 2.6 that contained , amongst other
            things, an offset patch. I guess this is what it was about.

            Thanks for the warning.
          • janne.kario
            ... Any progress on this? I grabbed yaffs sources from openwrt project. They also have a patch to make yaffs compile on 2.6.25+. I ve managed to get my
            Message 5 of 6 , Jun 5, 2008
            View Source
            • 0 Attachment
              --- In ts-7000@yahoogroups.com, "j.chitte" <j.chitte@...> wrote:
              >

              >
              > > If I rewrite the filesystem while in 2.6, it works fine, but when I
              > boot
              > > back into 2.4, it again shows a blank filesystem.
              > >

              Any progress on this? I grabbed yaffs sources from openwrt project.
              They also have a patch to make yaffs compile on 2.6.25+. I've managed
              to get my original tslinux 2.4 rootfs mounted (read-only) but I'm
              hesistant to try rw mode because I don't want to mess up anything.
              Also, yaffs is running on top of Alexander Clouters new NAND driver
              which I don't trust yet ;)

              j
            • Alexander Clouter
              Hi, ... Cheeky :) To be honest, it is actually the *same* driver just re-organised. The only real change between the ts7250.c/ts7xxx.c is that the partition
              Message 6 of 6 , Jun 5, 2008
              View Source
              • 0 Attachment
                Hi,

                janne.kario <janne.kario@...> [20080605 21:56:21 -0000]:
                >
                > Also, yaffs is running on top of Alexander Clouters new NAND driver
                > which I don't trust yet ;)
                >
                Cheeky :)

                To be honest, it is actually the *same* driver just re-organised. The only
                real change between the ts7250.c/ts7xxx.c is that the partition table
                decision making is moved from the NAND driver into
                arch/arm/mach-(ep93xx|orion5x)/ and made a runtime decision depending on the
                NAND size[1].

                The only thing that should screw up is the partition tabling...should :)

                Having said them once I have gotten the ts78xx code and ts7xxx NAND code
                mainlined I can look at why with DMA enabled things go horribly wrong...

                Cheers

                Alex

                [1] also the IO ports are also choosen in the platform code rather than in
                the NAND driver

                --
                _________________________________________
                < You buttered your bread, now lie in it. >
                -----------------------------------------
                \ ^__^
                \ (oo)\_______
                (__)\ )\/\
                ||----w |
                || ||
              Your message has been successfully submitted and would be delivered to recipients shortly.