Message ID | 20200213182950.10744-1-patrick.delaunay@st.com |
---|---|
State | New |
Headers | show |
Series | ARM: bootm: take into account gd->ram_top | expand |
Hi Patrick, On Thu, 13 Feb 2020 at 11:30, Patrick Delaunay <patrick.delaunay at st.com> wrote: > > From: Patrice Chotard <patrice.chotard at st.com> > > If gd->ram_top has been tuned using board_get_usable_ram_top(), > it must be taken into account when reserving arch lmb. > > Signed-off-by: Patrice Chotard <patrice.chotard at st.com> > Reviewed-by: Patrick DELAUNAY <patrick.delaunay at st.com> > Signed-off-by: Patrick Delaunay <patrick.delaunay at st.com> > --- > > arch/arm/lib/bootm.c | 3 +++ > 1 file changed, 3 insertions(+) Is this something we can test in test/lib/lmb.c ? Regards, Simon
Hi Simon, > From: Simon Glass <sjg at chromium.org> > Sent: vendredi 14 f?vrier 2020 20:17 > > Hi Patrick, > > On Thu, 13 Feb 2020 at 11:30, Patrick Delaunay <patrick.delaunay at st.com> > wrote: > > > > From: Patrice Chotard <patrice.chotard at st.com> > > > > If gd->ram_top has been tuned using board_get_usable_ram_top(), it > > must be taken into account when reserving arch lmb. > > > > Signed-off-by: Patrice Chotard <patrice.chotard at st.com> > > Reviewed-by: Patrick DELAUNAY <patrick.delaunay at st.com> > > Signed-off-by: Patrick Delaunay <patrick.delaunay at st.com> > > --- > > > > arch/arm/lib/bootm.c | 3 +++ > > 1 file changed, 3 insertions(+) > > Is this something we can test in test/lib/lmb.c ? I check these tests, and for me it is not possible, as I change the ARM part for bootm lib (limit the reserved size by u-boot to avoid conflict wxith other reserved memory) and not the lmb generic code... I don't see how to test it in sandbox. > Regards, > Simon Regards Patrick
On Thu, Feb 13, 2020 at 07:29:50PM +0100, Patrick Delaunay wrote: > From: Patrice Chotard <patrice.chotard at st.com> > > If gd->ram_top has been tuned using board_get_usable_ram_top(), > it must be taken into account when reserving arch lmb. > > Signed-off-by: Patrice Chotard <patrice.chotard at st.com> > Reviewed-by: Patrick DELAUNAY <patrick.delaunay at st.com> > Signed-off-by: Patrick Delaunay <patrick.delaunay at st.com> Applied to u-boot/master, thanks!
Hi Patrick, all, On Thu, Feb 13 2020, Patrick Delaunay wrote: > From: Patrice Chotard <patrice.chotard@st.com> > > If gd->ram_top has been tuned using board_get_usable_ram_top(), > it must be taken into account when reserving arch lmb. > > Signed-off-by: Patrice Chotard <patrice.chotard@st.com> > Reviewed-by: Patrick DELAUNAY <patrick.delaunay@st.com> > Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com> [snip] > diff --git a/arch/arm/lib/bootm.c b/arch/arm/lib/bootm.c > index a135bcfc7b..f4b5ca6de0 100644 > --- a/arch/arm/lib/bootm.c > +++ b/arch/arm/lib/bootm.c > @@ -75,6 +75,9 @@ void arch_lmb_reserve(struct lmb *lmb) > gd->bd->bi_dram[bank].size - 1; > if (sp > bank_end) > continue; > + if (bank_end > gd->ram_top) > + bank_end = gd->ram_top - 1; > + This patch (now committed as 8ce1f10cf2b1) breaks kernel boot on Armada 8040 based Clearfog GT-8K with 16GB RAM. See below the console output of v2020.10-rc2 with a few added prints. The first memory bank (bi_dram[0]) goes from 0 to 3GB. The rest (4GB-17GB) is on bi_dram[1] (see a8k_dram_init_banksize()). ram_top is set to 2GB on arch/arm/mach-mvebu/arm64-common.c:board_get_usable_ram_top(). Reverting commit 8ce1f10cf2b1 on top of v2020.10-rc2 fixes boot. Any Idea? Found /extlinux/extlinux.conf Retrieving file: /extlinux/extlinux.conf ## Current stack ends at 0x7fb24300 arch_lmb_reserve: bank_end: bfffffff ram_top: 80000000 62 bytes read in 21 ms (2 KiB/s) 1: linux Retrieving file: /extlinux/Image ## Current stack ends at 0x7fb23960 arch_lmb_reserve: bank_end: bfffffff ram_top: 80000000 13740544 bytes read in 1266 ms (10.4 MiB/s) Retrieving file: /extlinux/armada-8040-clearfog-gt-8k.dtb ## Current stack ends at 0x7fb23960 arch_lmb_reserve: bank_end: bfffffff ram_top: 80000000 33368 bytes read in 31 ms (1 MiB/s) ## Current stack ends at 0x7fb23cd0 arch_lmb_reserve: bank_end: bfffffff ram_top: 80000000 ## Flattened Device Tree blob at 04f00000 Booting using the fdt blob at 0x4f00000 Loading Device Tree to 00000000bfff4000, end 00000000bffff257 ... "Synchronous Abort" handler, esr 0x96000045 elr: 000000000006e1cc lr : 0000000000068fd8 (reloc) elr: 000000007ffa91cc lr : 000000007ffa3fd8 x0 : ffffffffffffffff x1 : 00000000bfffc258 x2 : 0000000000000000 x3 : ffffffffffff7da7 x4 : 0000000004f08258 x5 : 00000000bfff4000 x6 : 00000000bfff4000 x7 : 000000000000000f x8 : 000000007fb23bf8 x9 : 0000000000000008 x10: 00000000bffff257 x11: 00000000bffff257 x12: 0000000000000000 x13: fffffffffffff000 x14: 00000000bfff4000 x15: 0000000000000021 x16: 000000007ff7bc38 x17: 0000000000000000 x18: 000000007fb2add0 x19: 00000000bfff4000 x20: 0000000004f00000 x21: 000000000000b258 x22: 0000000058820000 x23: 0000000000000010 x24: 000000007ffe3c40 x25: 000000007fb23cb8 x26: 00000000c0000000 x27: 0000000000000000 x28: 000000007fc3fd50 x29: 000000007fb23bd0 Code: 54000061 aa0603e0 d65f03c0 38606882 (38206822) Resetting CPU ... Thanks, baruch > lmb_reserve(lmb, sp, bank_end - sp + 1); > break; > } -- ~. .~ Tk Open Systems =}------------------------------------------------ooO--U--Ooo------------{= - baruch@tkos.co.il - tel: +972.52.368.4656, http://www.tkos.co.il -
Hi Baruch On 8/17/20 6:46 AM, Baruch Siach wrote: > Hi Patrick, all, > > On Thu, Feb 13 2020, Patrick Delaunay wrote: >> From: Patrice Chotard <patrice.chotard@st.com> >> >> If gd->ram_top has been tuned using board_get_usable_ram_top(), >> it must be taken into account when reserving arch lmb. >> >> Signed-off-by: Patrice Chotard <patrice.chotard@st.com> >> Reviewed-by: Patrick DELAUNAY <patrick.delaunay@st.com> >> Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com> > [snip] >> diff --git a/arch/arm/lib/bootm.c b/arch/arm/lib/bootm.c >> index a135bcfc7b..f4b5ca6de0 100644 >> --- a/arch/arm/lib/bootm.c >> +++ b/arch/arm/lib/bootm.c >> @@ -75,6 +75,9 @@ void arch_lmb_reserve(struct lmb *lmb) >> gd->bd->bi_dram[bank].size - 1; >> if (sp > bank_end) >> continue; >> + if (bank_end > gd->ram_top) >> + bank_end = gd->ram_top - 1; >> + > This patch (now committed as 8ce1f10cf2b1) breaks kernel boot on Armada > 8040 based Clearfog GT-8K with 16GB RAM. See below the console output of > v2020.10-rc2 with a few added prints. > > The first memory bank (bi_dram[0]) goes from 0 to 3GB. The rest > (4GB-17GB) is on bi_dram[1] (see a8k_dram_init_banksize()). ram_top is > set to 2GB on > arch/arm/mach-mvebu/arm64-common.c:board_get_usable_ram_top(). If i have correctly understood, before this patch, on your platform U-boot was using a memory area over the ram_top limit set to 2Gb ? which is not correct. You probably have to rework the platform memory mapping in order to either _ make all needed memory area fit inside the ram_top limit (2Gb) or _ increase the ram_top limit over 2Gb. It's only assumption as i don't know your platform. Patrice > > Reverting commit 8ce1f10cf2b1 on top of v2020.10-rc2 fixes boot. > > Any Idea? > > Found /extlinux/extlinux.conf > Retrieving file: /extlinux/extlinux.conf > ## Current stack ends at 0x7fb24300 > arch_lmb_reserve: bank_end: bfffffff ram_top: 80000000 > 62 bytes read in 21 ms (2 KiB/s) > 1: linux > Retrieving file: /extlinux/Image > ## Current stack ends at 0x7fb23960 > arch_lmb_reserve: bank_end: bfffffff ram_top: 80000000 > 13740544 bytes read in 1266 ms (10.4 MiB/s) > Retrieving file: /extlinux/armada-8040-clearfog-gt-8k.dtb > ## Current stack ends at 0x7fb23960 > arch_lmb_reserve: bank_end: bfffffff ram_top: 80000000 > 33368 bytes read in 31 ms (1 MiB/s) > ## Current stack ends at 0x7fb23cd0 > arch_lmb_reserve: bank_end: bfffffff ram_top: 80000000 > ## Flattened Device Tree blob at 04f00000 > Booting using the fdt blob at 0x4f00000 > Loading Device Tree to 00000000bfff4000, end 00000000bffff257 ... "Synchronous Abort" handler, esr 0x96000045 > elr: 000000000006e1cc lr : 0000000000068fd8 (reloc) > elr: 000000007ffa91cc lr : 000000007ffa3fd8 > x0 : ffffffffffffffff x1 : 00000000bfffc258 > x2 : 0000000000000000 x3 : ffffffffffff7da7 > x4 : 0000000004f08258 x5 : 00000000bfff4000 > x6 : 00000000bfff4000 x7 : 000000000000000f > x8 : 000000007fb23bf8 x9 : 0000000000000008 > x10: 00000000bffff257 x11: 00000000bffff257 > x12: 0000000000000000 x13: fffffffffffff000 > x14: 00000000bfff4000 x15: 0000000000000021 > x16: 000000007ff7bc38 x17: 0000000000000000 > x18: 000000007fb2add0 x19: 00000000bfff4000 > x20: 0000000004f00000 x21: 000000000000b258 > x22: 0000000058820000 x23: 0000000000000010 > x24: 000000007ffe3c40 x25: 000000007fb23cb8 > x26: 00000000c0000000 x27: 0000000000000000 > x28: 000000007fc3fd50 x29: 000000007fb23bd0 > > Code: 54000061 aa0603e0 d65f03c0 38606882 (38206822) > Resetting CPU ... > > Thanks, > baruch > >> lmb_reserve(lmb, sp, bank_end - sp + 1); >> break; >> }
diff --git a/arch/arm/lib/bootm.c b/arch/arm/lib/bootm.c index a135bcfc7b..f4b5ca6de0 100644 --- a/arch/arm/lib/bootm.c +++ b/arch/arm/lib/bootm.c @@ -75,6 +75,9 @@ void arch_lmb_reserve(struct lmb *lmb) gd->bd->bi_dram[bank].size - 1; if (sp > bank_end) continue; + if (bank_end > gd->ram_top) + bank_end = gd->ram_top - 1; + lmb_reserve(lmb, sp, bank_end - sp + 1); break; }