Re: [PATCH RFC] x86 early_ioremap: increase FIX_BTMAPS_SLOTS to 8

[Dave Young <dyoung-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>]

On 09/14/14 at 03:14pm, Matt Fleming wrote:
> On Tue, 26 Aug, at 05:06:41PM, Dave Young wrote:
> > 3.16 kernel boot fail with earlyprintk=efi, it keeps scrolling at the
> > bottom line of screen.
> > 
> > Bisected, the first bad commit is below:
> > commit 86dfc6f339886559d80ee0d4bd20fe5ee90450f0
> > Author: Lv Zheng <lv.zheng-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
> > Date:   Fri Apr 4 12:38:57 2014 +0800
> > 
> >     ACPICA: Tables: Fix table checksums verification before installation.
> 
> [...]
> 
> > Quote reply from Lv.zheng about the early ioremap slot usage in this case:
> > 
> > """
> > In early_efi_scroll_up(), 2 mapping entries will be used for the src/dst screen buffer.
> > In drivers/acpi/acpica/tbutils.c, we've improved the early table loading code in acpi_tb_parse_root_table().
> > We now need 2 mapping entries:
> > 1. One mapping entry is used for RSDT table mapping. Each RSDT entry contains an address for another ACPI table.
> > 2. For each entry in RSDP, we need another mapping entry to map the table to perform necessary check/override before installing it.
> > 
> > When acpi_tb_parse_root_table() prints something through EFI earlyprintk console, we'll have 4 mapping entries used.
> > The current 4 slots setting of early_ioremap() seems to be too small for such a use case.
> > """
> > 
> > Thus increase the slot to 8 in this patch to fix this issue.
> > boot-time mappings become 512 page with this patch.
>  
> This analysis and solution makes sense to me, but I was really hoping
> for one of the x86 maintainers to ACK the below patch. I'm happy to take
> it through the EFI tree if it's not picked up into tip meanwhile.
> 
> Also, Dave, it sounds like this should be tagged for stable if the bug
> has been around since v3.16?

Matt, yes, it should be a stable issue, ccing Greg and stable list.

> 
> > Signed-off-by: Dave Young <dyoung-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
> > ---
> > I'm not sure if this is ok in 32bit, review and comments are appreciated.
> >  arch/x86/include/asm/fixmap.h |    6 +++---
> >  1 file changed, 3 insertions(+), 3 deletions(-)
> > 
> > Index: linux/arch/x86/include/asm/fixmap.h
> > ===================================================================
> > --- linux.orig/arch/x86/include/asm/fixmap.h
> > +++ linux/arch/x86/include/asm/fixmap.h
> > @@ -106,14 +106,14 @@ enum fixed_addresses {
> >  	__end_of_permanent_fixed_addresses,
> >  
> >  	/*
> > -	 * 256 temporary boot-time mappings, used by early_ioremap(),
> > +	 * 512 temporary boot-time mappings, used by early_ioremap(),
> >  	 * before ioremap() is functional.
> >  	 *
> > -	 * If necessary we round it up to the next 256 pages boundary so
> > +	 * If necessary we round it up to the next 512 pages boundary so
> >  	 * that we can have a single pgd entry and a single pte table:
> >  	 */
> >  #define NR_FIX_BTMAPS		64
> > -#define FIX_BTMAPS_SLOTS	4
> > +#define FIX_BTMAPS_SLOTS	8
> >  #define TOTAL_FIX_BTMAPS	(NR_FIX_BTMAPS * FIX_BTMAPS_SLOTS)
> >  	FIX_BTMAP_END =
> >  	 (__end_of_permanent_fixed_addresses ^
> 
> 
> -- 
> Matt Fleming, Intel Open Source Technology Center