On Sun, 11 Jan 2015 12:55 +0100
Lars Doelle <[email protected]> wrote:
> Hi everyone,
>
> the boot log shows a very unhappy nand driver.
>
> --
> [ 0.682894] Division by zero in kernel.
> [ 0.689714] [<c001540c>] (unwind_backtrace+0x0/0x12c) from [<c03d41f8>]
> (Ldiv0+0x8/0x10)
> [ 0.696123] [<c03d41f8>] (Ldiv0+0x8/0x10) from [<c047351c>]
> (get_cmu_clk+0x30/0x40)
> [ 0.703360] [<c047351c>] (get_cmu_clk+0x30/0x40) from [<c047354c>]
> (set_nand_clock+0x20/0x80)
> [ 0.710497] [<c047354c>] (set_nand_clock+0x20/0x80) from [<c09b2bb4>]
> (nand_init+0x68/0x1c4)
> [ 0.717906] [<c09b2bb4>] (nand_init+0x68/0x1c4) from [<c0008698>]
> (do_one_initcall+0x114/0x16c)
> [ 0.725577] [<c0008698>] (do_one_initcall+0x114/0x16c) from [<c09979f8>]
> (kernel_init+0x190/0x228)
> [ 0.733160] [<c09979f8>] (kernel_init+0x190/0x228) from [<c000f7ec>]
> (kernel_thread_exit+0x0/0x8)
> [ 0.736507] [NAND] nand driver version: 0x2 0x9
> [ 0.742146] [SCAN_ERR] search nand physical architecture parameter failed!
> [ 0.744506] [NAND]init_blklayer fail
> --
>
> Apparently, this happens in drivers/block/sunxi_nand/nfd/nand_blk.c:1107
>
> --
> #ifndef USE_SYS_CLK
> __u32 get_cmu_clk(void)
> {
> __u32 reg_val;
> __u32 div_p, factor_n;
> __u32 factor_k, factor_m;
> __u32 clock;
>
> reg_val = *(volatile unsigned int *)(0xf1c20000 + 0x20);
> div_p = (reg_val >> 16) & 0x3;
> factor_n = (reg_val >> 8) & 0x1f;
> factor_k = ((reg_val >> 4) & 0x3) + 1;
> factor_m = ((reg_val >> 0) & 0x3) + 1;
>
> clock = 24 * factor_n * factor_k/div_p/factor_m;
>
> return clock;
> }
> --
>
> From this code, since factor_m > 0, it is div_p being zero.
>
> The related Android log lines are:
> --
> <4>[ 1.651812] nand_init,line:1400
> <4>[ 1.651837] [NAND]nand driver, init.
> <4>[ 1.651852] [NAND] nand driver(A20) version: 0x2, 0x12, data: 20130627
> 1522
> <4>[ 1.651863] [NAND] nand clk request start
> <4>[ 1.651898] [NAND] nand clk request ok!
> <4>[ 1.651999] [NAND DMA] request dma success
> <4>[ 1.652011] [NAND DMA] set fulldone_cb success
> <4>[ 1.652410] nand interrupt register ok
> <4>[ 1.652426] [NAND] get nand_good_block_ratio from script: 870
> <4>[ 1.652506] [NAND DMA] start
> dma*****************************************
> --
>
> Since the driver mentions explicitly '(A20)', I might be running
> an unsuited driver. Is this a known A20 issue or particular to
> my installation? Instead using CONFIG_NAND, should I better
> give CONFIG_NAND_TEST a shot?
>
> The A20 handbook does not mention the register, but
>
> ./arch/arm/plat-sunxi/include/plat/platform.h:182 has a
> #define SW_VA_CCM_IO_BASE 0xf1c20000,
>
> and finally relates to CCM_PLL5_CFG this via
> http://linux-sunxi.org/Clock_Control_Module#Timer_Registers
> where it is cited as being A10 material.
>
> But even if it is A10-only material, something appears to be fishy
> either the code or the wiki page:
>
> | CCM_PLL5_P 16:17 Read/Write 0x00
> | 0x00 = 1
> | 0x01 = 2
> | 0x02 = 3
> | 0x03 = 4
> | PLL5 external factor P
>
> Here, 0x00 is perfectly legal and the above code should read IMO
>
> | div_p = ((reg_val >> 16) & 0x3) + 1;
>
> But perhaps I'm reading this wrong.
This is fixed in this commit:
https://github.com/linux-sunxi/linux-sunxi/commit/ade08aa6e5249a9e
To workaround problems in older sunxi-3.4 kernels, there is
CONFIG_OLD_SUNXI_KERNEL_COMPAT option in the mainline u-boot.
--
Best regards,
Siarhei Siamashka
--
You received this message because you are subscribed to the Google Groups
"linux-sunxi" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
