Re: [PATCH 1/1] board: sifive: unmatched: enlarge CONFIG_SYS_SPL_MALLOC_SIZE

[Rick Chen <rickchen36@gmail.com>]

> Hi Rick,
>
> On Mon, Oct 25, 2021 at 10:24 AM Rick Chen <rickchen36@gmail.com> wrote:
> >
> > Hi, Bin
> >
> > > Hi Rick,
> > >
> > > On Mon, Oct 25, 2021 at 9:49 AM Rick Chen <rickchen36@gmail.com> wrote:
> > > >
> > > > Hi Bin,
> > > >
> > > > > From: Bin Meng <bmeng.cn@gmail.com>
> > > > > Sent: Tuesday, October 19, 2021 4:55 PM
> > > > > To: Alexandre Ghiti <alexandre.ghiti@canonical.com>
> > > > > Cc: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>; Tom Rini <trini@konsulko.com>; Leo Yu-Chi Liang(梁育齊) <ycliang@andestech.com>; Rick Jian-Zhi Chen(陳建志) <rick@andestech.com>; Pragnesh Patel <pragnesh.patel@sifive.com>; Dimitri John Ledkov <dimitri.ledkov@canonical.com>; Zong Li <zong.li@sifive.com>; Green Wan <green.wan@sifive.com>; U-Boot Mailing List <u-boot@lists.denx.de>
> > > > > Subject: Re: [PATCH 1/1] board: sifive: unmatched: enlarge CONFIG_SYS_SPL_MALLOC_SIZE
> > > > >
> > > > > On Tue, Oct 19, 2021 at 4:32 PM Alexandre Ghiti <alexandre.ghiti@canonical.com> wrote:
> > > > > >
> > > > > > Hi,
> > > > > >
> > > > > > On Fri, Oct 1, 2021 at 5:35 PM Bin Meng <bmeng.cn@gmail.com> wrote:
> > > > > > >
> > > > > > > Hi Heinrich,
> > > > > > >
> > > > > > > On Fri, Oct 1, 2021 at 7:37 PM Heinrich Schuchardt
> > > > > > > <heinrich.schuchardt@canonical.com> wrote:
> > > > > > > >
> > > > > > > > Avoid an error like
> > > > > > > >
> > > > > > > >     Could not get FIT buffer of 1725952 bytes
> > > > > > > >             check CONFIG_SYS_SPL_MALLOC_SIZE
> > > > > > > >     No device tree specified in SPL image
> > > > > > > >     ### ERROR ### Please RESET the board ###
> > > > > > > >
> > > > > > > > Signed-off-by: Heinrich Schuchardt
> > > > > > > > <heinrich.schuchardt@canonical.com>
> > > > > > > > ---
> > > > > > > >  include/configs/sifive-unmatched.h | 2 +-
> > > > > > > >  1 file changed, 1 insertion(+), 1 deletion(-)
> > > > > > > >
> > > > > > > > diff --git a/include/configs/sifive-unmatched.h
> > > > > > > > b/include/configs/sifive-unmatched.h
> > > > > > > > index f8ad2cce1f..8d3deabdd3 100644
> > > > > > > > --- a/include/configs/sifive-unmatched.h
> > > > > > > > +++ b/include/configs/sifive-unmatched.h
> > > > > > > > @@ -18,7 +18,7 @@
> > > > > > > >  #define CONFIG_SPL_BSS_MAX_SIZE                0x00100000
> > > > > > > >  #define CONFIG_SYS_SPL_MALLOC_START    (CONFIG_SPL_BSS_START_ADDR + \
> > > > > > > >                                          CONFIG_SPL_BSS_MAX_SIZE)
> > > > > > > > -#define CONFIG_SYS_SPL_MALLOC_SIZE     0x00100000
> > > > > > > > +#define CONFIG_SYS_SPL_MALLOC_SIZE     0x00200000
> > > > > > > >
> > > > > > > >  #define CONFIG_SPL_STACK       (0x08000000 + 0x001D0000 - \
> > > > > > > >                                  GENERATED_GBL_DATA_SIZE)
> > > > > > >
> > > > > > > What caused this?
> > > > > > >
> > > > > > > Last time this was seen on Ax25-AE350, CONFIG_SPL_SYS_MALLOC_F_LEN
> > > > > > > was increased, instead of CONFIG_SYS_SPL_MALLOC_SIZE which the error
> > > > > > > messages point to
> > > > > > >
> > > > > > > https://lists.denx.de/pipermail/u-boot/2021-May/449447.html
> > > > > > >
> > > > > >
> > > > > > I fell into the same issue this morning and increasing
> > > > > > CONFIG_SYS_SPL_MALLOC_SIZE fixed it, though I had to increase it even
> > > > > > more than Heinrich.
> > > > >
> > > > > Is this default build that caused Unmatched boot failure?
> > > > >
> > > > > @Rick Chen  can you comment on why CONFIG_SPL_SYS_MALLOC_F_LEN was needed on AE350?
> > > >
> > > > It is needed for limited memory cases on AE350 platforms.
> > >
> > > But the error message indicates that CONFIG_SYS_SPL_MALLOC_SIZE should
> > > be increased, which is what this patch doing.
> > >
> > > Why is CONFIG_SYS_SPL_MALLOC_SIZE not working on AE350?
> >
> > I review why I increase SPL_SYS_MALLOC_F_LEN and recall that it will
> > report memory size problem in spl_get_fit_load_buffer() of spl_fit.c.
> > But Increase CONFIG_SYS_SPL_MALLOC_SIZE is helpful for boards with
> > CONFIG_SYS_SPL_MALLOC_SIZE definition,
> > On Ae350 it is not define SYS_SPL_MALLOC_SIZE, so I increase
> > SPL_SYS_MALLOC_F_LEN instead.
> >
>
> Thanks. So this suggests we should fix the message in
> spl_get_fit_load_buffer() to provide different hints for different
> configurations?

Yes, maybe we can add #if CONFIG_VAL(SYS_MALLOC_F_LEN) to distinguish
CONFIG_SYS_SPL_MALLOC_SIZE and get better hints here.

Thanks,
Rick

>
> Regards,
> Bin