Hi Igor,
According to my understanding to jansson implementation, json_object_seed() is
only called in three places. Two of them are in test code and we don't use them
in EDK2. The only chance to call json_object_seed() is in json_object(). So, it
is called when driver create new JSON object and hashtable_seed is not
initialized yet, which means json_object_seed() will be called only once per
UEFI driver.
json_object_seed() use current time as hash table seed, which creates issue in
EDK2 environment. In EDK2, library is built with UEFI driver together. So,
there are multiple instances of JsonLib in the system. And the hash table seed
is different in each driver in current implementation. As the result, the JSON
object created by this driver can not be used by another driver.
> Does that mean, that json_object_seed function is useless anymore with those
> changes?
Yes, once constructor set hashtable_seed to a magic value, json_object_seed()
won't do anything.
> And there is no need for other module to call json_object_seed?
No, it is called when driver create JSON object and the seed is still zero.
Once the seed is set to non-zero value, json_object_seed() does nothing.
Regards,
Nickle
> -----Original Message-----
> From: Igor Kulchytskyy <ig...@ami.com>
> Sent: Monday, September 18, 2023 9:42 PM
> To: Nickle Wang <nick...@nvidia.com>; devel@edk2.groups.io
> Cc: Abner Chang <abner.ch...@amd.com>; Nick Ramirez
> <nrami...@nvidia.com>
> Subject: RE: [EXTERNAL] [PATCH] RedfishPkg/JsonLib: fix JsonObjectGetValue
> issue
>
> External email: Use caution opening links or attachments
>
>
> Hi Nickle,
> Does that mean, that json_object_seed function is useless anymore with those
> changes?
> And there is no need for other module to call json_object_seed?
> Thank you,
> Igor
>
> -----Original Message-----
> From: Nickle Wang <nick...@nvidia.com>
> Sent: Monday, September 18, 2023 9:09 AM
> To: devel@edk2.groups.io
> Cc: Abner Chang <abner.ch...@amd.com>; Igor Kulchytskyy <ig...@ami.com>;
> Nick Ramirez <nrami...@nvidia.com>
> Subject: [EXTERNAL] [PATCH] RedfishPkg/JsonLib: fix JsonObjectGetValue issue
>
>
> **CAUTION: The e-mail below is from an external source. Please exercise
> caution before opening attachments, clicking links, or following guidance.**
>
> JsonObjectGetValue() cannot find corresponding JSON value when the
> EDKII_JSON_VALUE object is created by another UEFI driver. This is because
> "hashtable_seed" is initialized by current time while JsonLib is loaded. So,
> "hashtable_seed"
> will be different in each individual UEFI driver.
>
> Signed-off-by: Nickle Wang <nick...@nvidia.com>
> Cc: Abner Chang <abner.ch...@amd.com>
> Cc: Igor Kulchytskyy <ig...@ami.com>
> Cc: Nick Ramirez <nrami...@nvidia.com>
> ---
> RedfishPkg/Library/JsonLib/JsonLib.inf | 1 +
> RedfishPkg/Library/JsonLib/JsonLib.c | 34 ++++++++++++++++++++++++++
> 2 files changed, 35 insertions(+)
>
> diff --git a/RedfishPkg/Library/JsonLib/JsonLib.inf
> b/RedfishPkg/Library/JsonLib/JsonLib.inf
> index 9d52a622e13a..b0d1bcac7cb3 100644
> --- a/RedfishPkg/Library/JsonLib/JsonLib.inf
> +++ b/RedfishPkg/Library/JsonLib/JsonLib.inf
> @@ -15,6 +15,7 @@
> MODULE_TYPE = DXE_DRIVER
> VERSION_STRING = 1.0
> LIBRARY_CLASS = JsonLib|DXE_DRIVER UEFI_APPLICATION
> UEFI_DRIVER
> + CONSTRUCTOR = JsonLibConstructor
>
> #
> # VALID_ARCHITECTURES = IA32 X64 ARM AARCH64 RISCV64
> diff --git a/RedfishPkg/Library/JsonLib/JsonLib.c
> b/RedfishPkg/Library/JsonLib/JsonLib.c
> index 9b758b940293..a7ec2ab217a6 100644
> --- a/RedfishPkg/Library/JsonLib/JsonLib.c
> +++ b/RedfishPkg/Library/JsonLib/JsonLib.c
> @@ -19,6 +19,8 @@
>
> #include "jansson.h"
>
> +extern volatile UINT32 hashtable_seed;
> +
> /**
> The function is used to initialize a JSON value which contains a new JSON
> array,
> or NULL on error. Initially, the array is empty.
> @@ -1138,3 +1140,35 @@ JsonGetType (
> {
> return (EDKII_JSON_TYPE)(((json_t *)JsonValue)->type); }
> +
> +/**
> + JSON Library constructor.
> +
> + @param ImageHandle The image handle.
> + @param SystemTable The system table.
> +
> + @retval EFI_SUCCESS Protocol listener is registered successfully.
> +
> +**/
> +EFI_STATUS
> +EFIAPI
> +JsonLibConstructor (
> + IN EFI_HANDLE ImageHandle,
> + IN EFI_SYSTEM_TABLE *SystemTable
> + )
> +{
> + //
> + // hashtable_seed is initalized by current time while JsonLib is loaded.
> + // Due to above mechanism, hashtable_seed will be different in each
> +individual
> + // UEFI driver. As the result, the hash of same key in different UEFI
> +driver
> + // would be different. This breaks JsonObjectGetValue() because
> + // JsonObjectGetValue() won't be able to find corresponding JSON
> +value if
> + // this EDKII_JSON_VALUE is created by another UEFI driver.
> + //
> + // Initial the seed to a fixed magic value for JsonLib to be working
> +in all
> + // UEFI drivers.
> + //
> + hashtable_seed = 0xFDAE2143;
> +
> + return EFI_SUCCESS;
> +}
> --
> 2.17.1
>
> -The information contained in this message may be confidential and proprietary
> to American Megatrends (AMI). This communication is intended to be read only
> by
> the individual or entity to whom it is addressed or by their designee. If the
> reader
> of this message is not the intended recipient, you are on notice that any
> distribution of this message, in any form, is strictly prohibited. Please
> promptly
> notify the sender by reply e-mail or by telephone at 770-246-8600, and then
> delete or destroy all copies of the transmission.
-=-=-=-=-=-=-=-=-=-=-=-
Groups.io Links: You receive all messages sent to this group.
View/Reply Online (#108904): https://edk2.groups.io/g/devel/message/108904
Mute This Topic: https://groups.io/mt/101432663/21656
Group Owner: devel+ow...@edk2.groups.io
Unsubscribe: https://edk2.groups.io/g/devel/unsub [arch...@mail-archive.com]
-=-=-=-=-=-=-=-=-=-=-=-
