Re: [PATCH v2] MdeModulePkg/Core: Fix MaxAddress in Gcd

["Zeng, Star" <star.zeng@intel.com>]

Glad to know you found the way. :)

Thanks,
Star

-----Original Message-----
From: Roman Bacik [mailto:roman.bacik@broadcom.com] 
Sent: Friday, May 4, 2018 12:12 AM
To: Zeng, Star <star.zeng@intel.com>; edk2-devel@lists.01.org
Cc: Ni, Ruiyu <ruiyu.ni@intel.com>; Vladimir Olovyannikov <vladimir.olovyannikov@broadcom.com>
Subject: RE: [edk2] [PATCH v2] MdeModulePkg/Core: Fix MaxAddress in Gcd

Hi Star,

It turns out we can use EfiGcdAllocateAnySearchBottomUp instead of EfiGcdAllocateMaxAddressSearchBottomUp after few changes to our code.
Thank you very much for clarification of our misunderstanding. We withdraw the patch and will close the associated bug ticket.
Thanks,

Roman

> -----Original Message-----
> From: Roman Bacik [mailto:roman.bacik@broadcom.com]
> Sent: Thursday, May 3, 2018 7:47 AM
> To: 'Zeng, Star'; 'edk2-devel@lists.01.org'
> Cc: 'Ni, Ruiyu'; Vladimir Olovyannikov
> Subject: RE: [edk2] [PATCH v2] MdeModulePkg/Core: Fix MaxAddress in 
> Gcd
>
> Hi Star,
>
> Thank you very much for your reply. Since the BaseAddress is the 
> lowest address in our case we thought this is a bug and the bottom up 
> search
starts
> at BaseAddress and ends at BaseAddress+length-1. What would be the 
> best way to do a search starting from the BaseAddress as the lowest? 
> Do we
need
> to add another search type?
> Thanks,
>
> Roman
>
> -----Original Message-----
> From: Zeng, Star [mailto:star.zeng@intel.com]
> Sent: Wednesday, May 2, 2018 6:35 PM
> To: Roman Bacik; edk2-devel@lists.01.org
> Cc: Ni, Ruiyu; Vladimir Olovyannikov; Zeng, Star
> Subject: RE: [edk2] [PATCH v2] MdeModulePkg/Core: Fix MaxAddress in 
> Gcd
>
> The GcdAllocateType are EfiGcdAllocateMaxAddress*, the MaxAddress 
> comes from input BaseAddress parameter.
>
> The original code logic is correct according to PI spec.
>
> PI Spec:
> If GcdAllocateType is EfiGcdAllocateMaxAddressSearchBottomUp, then the 
> GCD memory space map is searched from the lowest address up to 
> BaseAddress looking for unallocated memory ranges of Length bytes 
> beginning on a boundary specified by Alignment that matches 
> GcdMemoryType.
> If GcdAllocateType is EfiGcdAllocateMaxAddressSearchTopDown, then the 
> GCD memory space map is searched from BaseAddress down to the lowest 
> address looking for unallocated memory ranges of Length bytes 
> beginning
on
> a boundary specified by Alignment that matches GcdMemoryType.
>
>
> Thanks,
> Star
> -----Original Message-----
> From: edk2-devel [mailto:edk2-devel-bounces@lists.01.org] On Behalf Of 
> Roman Bacik
> Sent: Thursday, May 3, 2018 2:01 AM
> To: edk2-devel@lists.01.org
> Cc: Ni, Ruiyu <ruiyu.ni@intel.com>; Vladimir Olovyannikov 
> <vladimir.olovyannikov@broadcom.com>
> Subject: [edk2] [PATCH v2] MdeModulePkg/Core: Fix MaxAddress in Gcd
>
> When BottomUp search is used the MaxAddress is incorrectly chosen to 
> be the BaseAddress instead of the EndAddress.
>
> Cc: Ruiyu Ni <ruiyu.ni@intel.com>
> Cc: Vladimir Olovyannikov <vladimir.olovyannikov@broadcom.com>
> Contributed-under: TianoCore Contribution Agreement 1.1
> Signed-off-by: Roman Bacik <roman.bacik@broadcom.com>
> ---
> MdeModulePkg/Core/Dxe/Gcd/Gcd.c | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/MdeModulePkg/Core/Dxe/Gcd/Gcd.c 
> b/MdeModulePkg/Core/Dxe/Gcd/Gcd.c index
> e17e98230b79..9eeb2bd74599 100644
> --- a/MdeModulePkg/Core/Dxe/Gcd/Gcd.c
> +++ b/MdeModulePkg/Core/Dxe/Gcd/Gcd.c
> @@ -1170,8 +1170,8 @@ CoreAllocateSpace ( // // Compute the maximum 
> address to use in the search algorithm //
> - if (GcdAllocateType == EfiGcdAllocateMaxAddressSearchBottomUp ||
> - GcdAllocateType == EfiGcdAllocateMaxAddressSearchTopDown ) {
> + if (GcdAllocateType == EfiGcdAllocateMaxAddressSearchTopDown || 
> + GcdAllocateType == EfiGcdAllocateAnySearchTopDown ) {
> MaxAddress = *BaseAddress;
> } else {
> MaxAddress = Entry->EndAddress;
> --
> 1.9.1
> _______________________________________________
> edk2-devel mailing list
> edk2-devel@lists.01.org
> https://lists.01.org/mailman/listinfo/edk2-devel