Re: [U-Boot] [PATCH 1/2] common: Generic file system firmware loader
On 11/16/2017 09:09 AM, Chee, Tien Fong wrote: > On Jum, 2017-11-10 at 11:04 +0100, Marek Vasut wrote: >> On 11/10/2017 10:05 AM, Chee, Tien Fong wrote: >>> >>> On Kha, 2017-11-09 at 11:31 +0100, Marek Vasut wrote: On 11/09/2017 11:00 AM, Lukasz Majewski wrote: > > > On Thu, 9 Nov 2017 08:05:18 +0100 > Marek Vasut wrote: > >> >> >> On 11/09/2017 07:04 AM, Chee, Tien Fong wrote: >>> >>> >>> On Sel, 2017-11-07 at 10:34 +0100, Marek Vasut wrote: On 11/07/2017 10:03 AM, Chee, Tien Fong wrote: > > > > On Isn, 2017-11-06 at 11:56 +0100, Marek Vasut wrote: >> >> >> >> On 11/06/2017 05:15 AM, Chee, Tien Fong wrote: >>> >>> >>> >>> >>> On Ahd, 2017-11-05 at 17:43 +0100, Marek Vasut >>> wrote: On 11/02/2017 09:20 AM, Chee, Tien Fong wrote: > > > > > > On Rab, 2017-11-01 at 10:26 +0100, Marek Vasut > wrote: >> >> >> >> >> >> On 11/01/2017 10:18 AM, tien.fong.chee@intel. >> com >> wrote: >>> >>> >>> >>> >>> >>> >>> From: Tien Fong Chee >> com> >>> >>> Generic firmware loader framework contains >>> some >>> common >>> functionality >>> which is factored out from splash loader. >>> It is >>> reusable by >>> any >>> specific driver file system firmware >>> loader. >>> Specific >>> driver >>> file >>> system >>> firmware loader handling can be defined >>> with >>> both weak >>> function >>> fsloader_preprocess and fs_loading. >>> >>> Signed-off-by: Tien Fong Chee >> ee@i >>> ntel.com >>> --- >>> common/Makefile | 1 + >>> common/load_fs.c | 217 >>> +++ >>> >>> +++ >>> include/load_fs.h | 38 ++ >>> 3 files changed, 256 insertions(+) >>> create mode 100644 common/load_fs.c >>> create mode 100644 include/load_fs.h >> [...] >> >>> >>> >>> >>> >>> >>> >>> +int flash_select_fs_dev(struct >>> flash_location >>> *location) >> Why does everything have flash_ prefix ? >> > I can remove the flash_ prefix, this generic FS > loader > should > support > for all filesystem instead of flash. > >> >> >> >> >> >> I also mentioned the API should copy the >> linux >> firmware >> loader >> API. >> > If i'm not mistaken, you are referring firmware > loader API > in > this > link https://github.com/torvalds/linux/blob/f00 > 7cad > 159e99fa > 2acd > 3b2e > 9364 > fbb32ad28b971/drivers/base/firmware_class.c#L12 > 64. > >>> I would like to confirm with you whether we are >>> talking >>> to the >>> same >>> API >>> above? >> https://www.kernel.org/doc/html/v4.13/driver-api/firm >> ware >> /index.h >> tml >> >> first link on google btw . You might be able to avoid >> the >> firmware >> structure. >> > After assessment, i found that Linux loader is not > suitable > for > fpga > loader as fpga loader need > 1) Able to program FPGA image in SPL chunk bu chunk > with > small > memory > allocatted. > 2) Name of FPGA image defined in DTS, and path of FPGA > image in > FAT and > UBI partition. > > Linux loader is strongly designed based on Linux > environment, > always > assume having RFF, env support(which SPL don't have), > sysfs > and > udev > feature. Sigh, you can just have some additional function call to fetch smaller chu
Re: [U-Boot] [PATCH 1/2] common: Generic file system firmware loader
On Jum, 2017-11-10 at 11:04 +0100, Marek Vasut wrote: > On 11/10/2017 10:05 AM, Chee, Tien Fong wrote: > > > > On Kha, 2017-11-09 at 11:31 +0100, Marek Vasut wrote: > > > > > > On 11/09/2017 11:00 AM, Lukasz Majewski wrote: > > > > > > > > > > > > On Thu, 9 Nov 2017 08:05:18 +0100 > > > > Marek Vasut wrote: > > > > > > > > > > > > > > > > > > > On 11/09/2017 07:04 AM, Chee, Tien Fong wrote: > > > > > > > > > > > > > > > > > > On Sel, 2017-11-07 at 10:34 +0100, Marek Vasut wrote: > > > > > > > > > > > > > > > > > > > > > On 11/07/2017 10:03 AM, Chee, Tien Fong wrote: > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > On Isn, 2017-11-06 at 11:56 +0100, Marek Vasut wrote: > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > On 11/06/2017 05:15 AM, Chee, Tien Fong wrote: > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > On Ahd, 2017-11-05 at 17:43 +0100, Marek Vasut > > > > > > > > > > wrote: > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > On 11/02/2017 09:20 AM, Chee, Tien Fong wrote: > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > On Rab, 2017-11-01 at 10:26 +0100, Marek Vasut > > > > > > > > > > > > wrote: > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > On 11/01/2017 10:18 AM, tien.fong.chee@intel. > > > > > > > > > > > > > com > > > > > > > > > > > > > wrote: > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > From: Tien Fong Chee > > > > > > > > > > > > > com> > > > > > > > > > > > > > > > > > > > > > > > > > > > > Generic firmware loader framework contains > > > > > > > > > > > > > > some > > > > > > > > > > > > > > common > > > > > > > > > > > > > > functionality > > > > > > > > > > > > > > which is factored out from splash loader. > > > > > > > > > > > > > > It is > > > > > > > > > > > > > > reusable by > > > > > > > > > > > > > > any > > > > > > > > > > > > > > specific driver file system firmware > > > > > > > > > > > > > > loader. > > > > > > > > > > > > > > Specific > > > > > > > > > > > > > > driver > > > > > > > > > > > > > > file > > > > > > > > > > > > > > system > > > > > > > > > > > > > > firmware loader handling can be defined > > > > > > > > > > > > > > with > > > > > > > > > > > > > > both weak > > > > > > > > > > > > > > function > > > > > > > > > > > > > > fsloader_preprocess and fs_loading. > > > > > > > > > > > > > > > > > > > > > > > > > > > > Signed-off-by: Tien Fong Chee > > > > > > > > > > > > > ee@i > > > > > > > > > > > > > > ntel.com > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > --- > > > > > > > > > > > > > > common/Makefile | 1 + > > > > > > > > > > > > > > common/load_fs.c | 217 > > > > > > > > > > > > > > +++ > > > > > > > > > > > > > > > > > > > > > > > > > > > > +++ > > > > > > > > > > > > > > include/load_fs.h | 38 ++ > > > > > > > > > > > > > > 3 files changed, 256 insertions(+) > > > > > > > > > > > > > > create mode 100644 common/load_fs.c > > > > > > > > > > > > > > create mode 100644 include/load_fs.h > > > > > > > > > > > > > [...] > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > +int flash_select_fs_dev(struct > > > > > > > > > > > > > > flash_location > > > > > > > > > > > > > > *location) > > > > > > > > > > > > > Why does everything have flash_ prefix ? > > > > > > > > > > > > > > > > > > > > > > > > > I can remove the flash_ prefix, this generic FS > > > > > > > > > > > > loader > > > > > > > > > > > > should > > > > > > > > > > > > support > > > > > > > > > > > > for all filesystem instead of flash. > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > I also mentioned the API should copy the > > > > > > > > > > > > > linux > > > > > > > > > > > > > firmware > > > > > > > > > > > > > loader > > > > > > > > > > > > > API. > > > > > > > > > > > > > > > > > > > > > > > > > If i'm not mistaken, you are referring firmware > > > > > > > > > > > > loader API > > > > > > > > > > > > in > > > > > > > > > > > > this > > > > > > > > > > > > link https://github.com/torvalds/linux/blob/f00 > > > > > > > > > > > > 7cad > > > > > > > > > > > > 159e99fa > > > > > > > > > >
Re: [U-Boot] [PATCH 1/2] common: Generic file system firmware loader
On Jum, 2017-11-10 at 11:04 +0100, Marek Vasut wrote: > On 11/10/2017 10:05 AM, Chee, Tien Fong wrote: > > > > On Kha, 2017-11-09 at 11:31 +0100, Marek Vasut wrote: > > > > > > On 11/09/2017 11:00 AM, Lukasz Majewski wrote: > > > > > > > > > > > > On Thu, 9 Nov 2017 08:05:18 +0100 > > > > Marek Vasut wrote: > > > > > > > > > > > > > > > > > > > On 11/09/2017 07:04 AM, Chee, Tien Fong wrote: > > > > > > > > > > > > > > > > > > On Sel, 2017-11-07 at 10:34 +0100, Marek Vasut wrote: > > > > > > > > > > > > > > > > > > > > > On 11/07/2017 10:03 AM, Chee, Tien Fong wrote: > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > On Isn, 2017-11-06 at 11:56 +0100, Marek Vasut wrote: > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > On 11/06/2017 05:15 AM, Chee, Tien Fong wrote: > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > On Ahd, 2017-11-05 at 17:43 +0100, Marek Vasut > > > > > > > > > > wrote: > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > On 11/02/2017 09:20 AM, Chee, Tien Fong wrote: > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > On Rab, 2017-11-01 at 10:26 +0100, Marek Vasut > > > > > > > > > > > > wrote: > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > On 11/01/2017 10:18 AM, tien.fong.chee@intel. > > > > > > > > > > > > > com > > > > > > > > > > > > > wrote: > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > From: Tien Fong Chee > > > > > > > > > > > > > com> > > > > > > > > > > > > > > > > > > > > > > > > > > > > Generic firmware loader framework contains > > > > > > > > > > > > > > some > > > > > > > > > > > > > > common > > > > > > > > > > > > > > functionality > > > > > > > > > > > > > > which is factored out from splash loader. > > > > > > > > > > > > > > It is > > > > > > > > > > > > > > reusable by > > > > > > > > > > > > > > any > > > > > > > > > > > > > > specific driver file system firmware > > > > > > > > > > > > > > loader. > > > > > > > > > > > > > > Specific > > > > > > > > > > > > > > driver > > > > > > > > > > > > > > file > > > > > > > > > > > > > > system > > > > > > > > > > > > > > firmware loader handling can be defined > > > > > > > > > > > > > > with > > > > > > > > > > > > > > both weak > > > > > > > > > > > > > > function > > > > > > > > > > > > > > fsloader_preprocess and fs_loading. > > > > > > > > > > > > > > > > > > > > > > > > > > > > Signed-off-by: Tien Fong Chee > > > > > > > > > > > > > ee@i > > > > > > > > > > > > > > ntel.com > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > --- > > > > > > > > > > > > > > common/Makefile | 1 + > > > > > > > > > > > > > > common/load_fs.c | 217 > > > > > > > > > > > > > > +++ > > > > > > > > > > > > > > > > > > > > > > > > > > > > +++ > > > > > > > > > > > > > > include/load_fs.h | 38 ++ > > > > > > > > > > > > > > 3 files changed, 256 insertions(+) > > > > > > > > > > > > > > create mode 100644 common/load_fs.c > > > > > > > > > > > > > > create mode 100644 include/load_fs.h > > > > > > > > > > > > > [...] > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > +int flash_select_fs_dev(struct > > > > > > > > > > > > > > flash_location > > > > > > > > > > > > > > *location) > > > > > > > > > > > > > Why does everything have flash_ prefix ? > > > > > > > > > > > > > > > > > > > > > > > > > I can remove the flash_ prefix, this generic FS > > > > > > > > > > > > loader > > > > > > > > > > > > should > > > > > > > > > > > > support > > > > > > > > > > > > for all filesystem instead of flash. > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > I also mentioned the API should copy the > > > > > > > > > > > > > linux > > > > > > > > > > > > > firmware > > > > > > > > > > > > > loader > > > > > > > > > > > > > API. > > > > > > > > > > > > > > > > > > > > > > > > > If i'm not mistaken, you are referring firmware > > > > > > > > > > > > loader API > > > > > > > > > > > > in > > > > > > > > > > > > this > > > > > > > > > > > > link https://github.com/torvalds/linux/blob/f00 > > > > > > > > > > > > 7cad > > > > > > > > > > > > 159e99fa > > > > > > > > > >
Re: [U-Boot] [PATCH 1/2] common: Generic file system firmware loader
On 11/10/2017 10:05 AM, Chee, Tien Fong wrote: > On Kha, 2017-11-09 at 11:31 +0100, Marek Vasut wrote: >> On 11/09/2017 11:00 AM, Lukasz Majewski wrote: >>> >>> On Thu, 9 Nov 2017 08:05:18 +0100 >>> Marek Vasut wrote: >>> On 11/09/2017 07:04 AM, Chee, Tien Fong wrote: > > On Sel, 2017-11-07 at 10:34 +0100, Marek Vasut wrote: >> >> On 11/07/2017 10:03 AM, Chee, Tien Fong wrote: >>> >>> >>> On Isn, 2017-11-06 at 11:56 +0100, Marek Vasut wrote: On 11/06/2017 05:15 AM, Chee, Tien Fong wrote: > > > > On Ahd, 2017-11-05 at 17:43 +0100, Marek Vasut wrote: >> >> >> >> On 11/02/2017 09:20 AM, Chee, Tien Fong wrote: >>> >>> >>> >>> >>> On Rab, 2017-11-01 at 10:26 +0100, Marek Vasut >>> wrote: On 11/01/2017 10:18 AM, tien.fong.c...@intel.com wrote: > > > > > > From: Tien Fong Chee > > Generic firmware loader framework contains some > common > functionality > which is factored out from splash loader. It is > reusable by > any > specific driver file system firmware loader. > Specific > driver > file > system > firmware loader handling can be defined with > both weak > function > fsloader_preprocess and fs_loading. > > Signed-off-by: Tien Fong Chee ntel.com >> >> > --- > common/Makefile | 1 + > common/load_fs.c | 217 > +++ > +++ > include/load_fs.h | 38 ++ > 3 files changed, 256 insertions(+) > create mode 100644 common/load_fs.c > create mode 100644 include/load_fs.h [...] > > > > > > +int flash_select_fs_dev(struct flash_location > *location) Why does everything have flash_ prefix ? >>> I can remove the flash_ prefix, this generic FS >>> loader >>> should >>> support >>> for all filesystem instead of flash. >>> I also mentioned the API should copy the linux firmware loader API. >>> If i'm not mistaken, you are referring firmware >>> loader API >>> in >>> this >>> link https://github.com/torvalds/linux/blob/f007cad >>> 159e99fa >>> 2acd >>> 3b2e >>> 9364 >>> fbb32ad28b971/drivers/base/firmware_class.c#L1264. >>> > I would like to confirm with you whether we are talking > to the > same > API > above? https://www.kernel.org/doc/html/v4.13/driver-api/firmware /index.h tml first link on google btw . You might be able to avoid the firmware structure. >>> After assessment, i found that Linux loader is not suitable >>> for >>> fpga >>> loader as fpga loader need >>> 1) Able to program FPGA image in SPL chunk bu chunk with >>> small >>> memory >>> allocatted. >>> 2) Name of FPGA image defined in DTS, and path of FPGA >>> image in >>> FAT and >>> UBI partition. >>> >>> Linux loader is strongly designed based on Linux >>> environment, >>> always >>> assume having RFF, env support(which SPL don't have), sysfs >>> and >>> udev >>> feature. >> Sigh, you can just have some additional function call to >> fetch >> smaller >> chunks from a file, I don't think it's that hard of a problem >> ... >> > We already have that function to support smaller chunks, and it > also > work for single large image when enough memory is available in > ver 1 > series of fpga loadfs. > > Since the Linux loader API is totally not suitable for fpga > loadfs, > and also nothing i can leverage from there for fpga loadfs, > could > you please consider to accept implementation for this series > patches or implementation for fpga loadfs series ver1? You mean going back to completely custom non-generic altera-only ad-hoc interface ? I'd like to hear opinion of the others on CC ... >>> I must admit that I don't know the exact Altera API for loading >>> their >>> bitstream. >> That's irrelevant for a generic loader. Th
Re: [U-Boot] [PATCH 1/2] common: Generic file system firmware loader
On Kha, 2017-11-09 at 11:31 +0100, Marek Vasut wrote: > On 11/09/2017 11:00 AM, Lukasz Majewski wrote: > > > > On Thu, 9 Nov 2017 08:05:18 +0100 > > Marek Vasut wrote: > > > > > > > > On 11/09/2017 07:04 AM, Chee, Tien Fong wrote: > > > > > > > > On Sel, 2017-11-07 at 10:34 +0100, Marek Vasut wrote: > > > > > > > > > > On 11/07/2017 10:03 AM, Chee, Tien Fong wrote: > > > > > > > > > > > > > > > > > > On Isn, 2017-11-06 at 11:56 +0100, Marek Vasut wrote: > > > > > > > > > > > > > > > > > > > > > On 11/06/2017 05:15 AM, Chee, Tien Fong wrote: > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > On Ahd, 2017-11-05 at 17:43 +0100, Marek Vasut wrote: > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > On 11/02/2017 09:20 AM, Chee, Tien Fong wrote: > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > On Rab, 2017-11-01 at 10:26 +0100, Marek Vasut > > > > > > > > > > wrote: > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > On 11/01/2017 10:18 AM, tien.fong.c...@intel.com > > > > > > > > > > > wrote: > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > From: Tien Fong Chee > > > > > > > > > > > > > > > > > > > > > > > > Generic firmware loader framework contains some > > > > > > > > > > > > common > > > > > > > > > > > > functionality > > > > > > > > > > > > which is factored out from splash loader. It is > > > > > > > > > > > > reusable by > > > > > > > > > > > > any > > > > > > > > > > > > specific driver file system firmware loader. > > > > > > > > > > > > Specific > > > > > > > > > > > > driver > > > > > > > > > > > > file > > > > > > > > > > > > system > > > > > > > > > > > > firmware loader handling can be defined with > > > > > > > > > > > > both weak > > > > > > > > > > > > function > > > > > > > > > > > > fsloader_preprocess and fs_loading. > > > > > > > > > > > > > > > > > > > > > > > > Signed-off-by: Tien Fong Chee > > > > > > > > > > > ntel.com > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > --- > > > > > > > > > > > > common/Makefile | 1 + > > > > > > > > > > > > common/load_fs.c | 217 > > > > > > > > > > > > +++ > > > > > > > > > > > > +++ > > > > > > > > > > > > include/load_fs.h | 38 ++ > > > > > > > > > > > > 3 files changed, 256 insertions(+) > > > > > > > > > > > > create mode 100644 common/load_fs.c > > > > > > > > > > > > create mode 100644 include/load_fs.h > > > > > > > > > > > [...] > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > +int flash_select_fs_dev(struct flash_location > > > > > > > > > > > > *location) > > > > > > > > > > > Why does everything have flash_ prefix ? > > > > > > > > > > > > > > > > > > > > > I can remove the flash_ prefix, this generic FS > > > > > > > > > > loader > > > > > > > > > > should > > > > > > > > > > support > > > > > > > > > > for all filesystem instead of flash. > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > I also mentioned the API should copy the linux > > > > > > > > > > > firmware > > > > > > > > > > > loader > > > > > > > > > > > API. > > > > > > > > > > > > > > > > > > > > > If i'm not mistaken, you are referring firmware > > > > > > > > > > loader API > > > > > > > > > > in > > > > > > > > > > this > > > > > > > > > > link https://github.com/torvalds/linux/blob/f007cad > > > > > > > > > > 159e99fa > > > > > > > > > > 2acd > > > > > > > > > > 3b2e > > > > > > > > > > 9364 > > > > > > > > > > fbb32ad28b971/drivers/base/firmware_class.c#L1264. > > > > > > > > > > > > > > > > > > I would like to confirm with you whether we are talking > > > > > > > > to the > > > > > > > > same > > > > > > > > API > > > > > > > > above? > > > > > > > https://www.kernel.org/doc/html/v4.13/driver-api/firmware > > > > > > > /index.h > > > > > > > tml > > > > > > > > > > > > > > first link on google btw . You might be able to avoid the > > > > > > > firmware > > > > > > > structure. > > > > > > > > > > > > > After assessment, i found that Linux loader is not suitable > > > > > > for > > > > > > fpga > > > > > > loader as fpga loader need > > > > > > 1) Able to program FPGA image in SPL chunk bu chunk with > > > > > > small > > > > > > memory > > > > > > allocatted. > > > > > > 2) Name of FPGA image defined in DTS, and path of FPGA > > > > > > image in > > > > > > FAT and > > > > > > UBI partition. > > > > > > > > > > > > Linux loader is strongly designed based on Linux > > > > > > environment, > > > > > > always > > > > > > assume having RFF, env support(which SPL don't have),
Re: [U-Boot] [PATCH 1/2] common: Generic file system firmware loader
On 11/09/2017 11:00 AM, Lukasz Majewski wrote: > On Thu, 9 Nov 2017 08:05:18 +0100 > Marek Vasut wrote: > >> On 11/09/2017 07:04 AM, Chee, Tien Fong wrote: >>> On Sel, 2017-11-07 at 10:34 +0100, Marek Vasut wrote: On 11/07/2017 10:03 AM, Chee, Tien Fong wrote: > > On Isn, 2017-11-06 at 11:56 +0100, Marek Vasut wrote: >> >> On 11/06/2017 05:15 AM, Chee, Tien Fong wrote: >>> >>> >>> On Ahd, 2017-11-05 at 17:43 +0100, Marek Vasut wrote: On 11/02/2017 09:20 AM, Chee, Tien Fong wrote: > > > > On Rab, 2017-11-01 at 10:26 +0100, Marek Vasut wrote: >> >> >> >> On 11/01/2017 10:18 AM, tien.fong.c...@intel.com wrote: >>> >>> >>> >>> >>> From: Tien Fong Chee >>> >>> Generic firmware loader framework contains some common >>> functionality >>> which is factored out from splash loader. It is >>> reusable by >>> any >>> specific driver file system firmware loader. Specific >>> driver >>> file >>> system >>> firmware loader handling can be defined with both weak >>> function >>> fsloader_preprocess and fs_loading. >>> >>> Signed-off-by: Tien Fong Chee >>> >>> --- >>> common/Makefile | 1 + >>> common/load_fs.c | 217 >>> ++ >>> include/load_fs.h | 38 ++ >>> 3 files changed, 256 insertions(+) >>> create mode 100644 common/load_fs.c >>> create mode 100644 include/load_fs.h >> [...] >> >>> >>> >>> >>> >>> +int flash_select_fs_dev(struct flash_location >>> *location) >> Why does everything have flash_ prefix ? >> > I can remove the flash_ prefix, this generic FS loader > should > support > for all filesystem instead of flash. > >> >> >> >> I also mentioned the API should copy the linux firmware >> loader >> API. >> > If i'm not mistaken, you are referring firmware loader API > in > this > link https://github.com/torvalds/linux/blob/f007cad159e99fa > 2acd > 3b2e > 9364 > fbb32ad28b971/drivers/base/firmware_class.c#L1264. > >>> I would like to confirm with you whether we are talking to the >>> same >>> API >>> above? >> https://www.kernel.org/doc/html/v4.13/driver-api/firmware/index.h >> tml >> >> first link on google btw . You might be able to avoid the >> firmware >> structure. >> > After assessment, i found that Linux loader is not suitable for > fpga > loader as fpga loader need > 1) Able to program FPGA image in SPL chunk bu chunk with small > memory > allocatted. > 2) Name of FPGA image defined in DTS, and path of FPGA image in > FAT and > UBI partition. > > Linux loader is strongly designed based on Linux environment, > always > assume having RFF, env support(which SPL don't have), sysfs and > udev > feature. Sigh, you can just have some additional function call to fetch smaller chunks from a file, I don't think it's that hard of a problem ... >>> We already have that function to support smaller chunks, and it also >>> work for single large image when enough memory is available in ver 1 >>> series of fpga loadfs. >>> >>> Since the Linux loader API is totally not suitable for fpga loadfs, >>> and also nothing i can leverage from there for fpga loadfs, could >>> you please consider to accept implementation for this series >>> patches or implementation for fpga loadfs series ver1? >> >> You mean going back to completely custom non-generic altera-only >> ad-hoc interface ? I'd like to hear opinion of the others on CC ... >> > > I must admit that I don't know the exact Altera API for loading their > bitstream. That's irrelevant for a generic loader. The loader should provide a file or ability to read chunks of file if needed, that's all. The consumer driver would then use that API to program whatever, ie. the FPGA. > What I would like to have though is a some kind of generic code, which > would allow me to reuse it on other ARM + DSP SoCs. ... on other platforms in general. > If we cannot re-use Linux stuff, then when we add something different > (more customer/industry aligned), please make it reusable for other > solutions (Xilinx, ADI, etc) - that would require a good documentation. And what is the problem with the Linux API ? I am not saying to reuse the Linux code, but the API is quite well fleshed out. -- Best regards, Marek Vasut ___ U-Bo
Re: [U-Boot] [PATCH 1/2] common: Generic file system firmware loader
On Thu, 9 Nov 2017 08:05:18 +0100 Marek Vasut wrote: > On 11/09/2017 07:04 AM, Chee, Tien Fong wrote: > > On Sel, 2017-11-07 at 10:34 +0100, Marek Vasut wrote: > >> On 11/07/2017 10:03 AM, Chee, Tien Fong wrote: > >>> > >>> On Isn, 2017-11-06 at 11:56 +0100, Marek Vasut wrote: > > On 11/06/2017 05:15 AM, Chee, Tien Fong wrote: > > > > > > On Ahd, 2017-11-05 at 17:43 +0100, Marek Vasut wrote: > >> > >> > >> On 11/02/2017 09:20 AM, Chee, Tien Fong wrote: > >>> > >>> > >>> > >>> On Rab, 2017-11-01 at 10:26 +0100, Marek Vasut wrote: > > > > On 11/01/2017 10:18 AM, tien.fong.c...@intel.com wrote: > > > > > > > > > > From: Tien Fong Chee > > > > Generic firmware loader framework contains some common > > functionality > > which is factored out from splash loader. It is > > reusable by > > any > > specific driver file system firmware loader. Specific > > driver > > file > > system > > firmware loader handling can be defined with both weak > > function > > fsloader_preprocess and fs_loading. > > > > Signed-off-by: Tien Fong Chee >> > > --- > > common/Makefile | 1 + > > common/load_fs.c | 217 > > ++ > > include/load_fs.h | 38 ++ > > 3 files changed, 256 insertions(+) > > create mode 100644 common/load_fs.c > > create mode 100644 include/load_fs.h > [...] > > > > > > > > > > > +int flash_select_fs_dev(struct flash_location > > *location) > Why does everything have flash_ prefix ? > > >>> I can remove the flash_ prefix, this generic FS loader > >>> should > >>> support > >>> for all filesystem instead of flash. > >>> > > > > I also mentioned the API should copy the linux firmware > loader > API. > > >>> If i'm not mistaken, you are referring firmware loader API > >>> in > >>> this > >>> link https://github.com/torvalds/linux/blob/f007cad159e99fa > >>> 2acd > >>> 3b2e > >>> 9364 > >>> fbb32ad28b971/drivers/base/firmware_class.c#L1264. > >>> > > I would like to confirm with you whether we are talking to the > > same > > API > > above? > https://www.kernel.org/doc/html/v4.13/driver-api/firmware/index.h > tml > > first link on google btw . You might be able to avoid the > firmware > structure. > > >>> After assessment, i found that Linux loader is not suitable for > >>> fpga > >>> loader as fpga loader need > >>> 1) Able to program FPGA image in SPL chunk bu chunk with small > >>> memory > >>> allocatted. > >>> 2) Name of FPGA image defined in DTS, and path of FPGA image in > >>> FAT and > >>> UBI partition. > >>> > >>> Linux loader is strongly designed based on Linux environment, > >>> always > >>> assume having RFF, env support(which SPL don't have), sysfs and > >>> udev > >>> feature. > >> Sigh, you can just have some additional function call to fetch > >> smaller > >> chunks from a file, I don't think it's that hard of a problem ... > >> > > We already have that function to support smaller chunks, and it also > > work for single large image when enough memory is available in ver 1 > > series of fpga loadfs. > > > > Since the Linux loader API is totally not suitable for fpga loadfs, > > and also nothing i can leverage from there for fpga loadfs, could > > you please consider to accept implementation for this series > > patches or implementation for fpga loadfs series ver1? > > You mean going back to completely custom non-generic altera-only > ad-hoc interface ? I'd like to hear opinion of the others on CC ... > I must admit that I don't know the exact Altera API for loading their bitstream. What I would like to have though is a some kind of generic code, which would allow me to reuse it on other ARM + DSP SoCs. If we cannot re-use Linux stuff, then when we add something different (more customer/industry aligned), please make it reusable for other solutions (Xilinx, ADI, etc) - that would require a good documentation. Those are my 2 cents. Best regards, Lukasz Majewski -- DENX Software Engineering GmbH, Managing Director: Wolfgang Denk HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: w...@denx.de pgpd0XD3M1Blk.pgp Description: OpenPGP digital signature ___ U-Boot mailing list U-Boot@lists.denx.de https://lists.denx.de/listinfo/u-boot
Re: [U-Boot] [PATCH 1/2] common: Generic file system firmware loader
On Kha, 2017-11-09 at 08:05 +0100, Marek Vasut wrote: > On 11/09/2017 07:04 AM, Chee, Tien Fong wrote: > > > > On Sel, 2017-11-07 at 10:34 +0100, Marek Vasut wrote: > > > > > > On 11/07/2017 10:03 AM, Chee, Tien Fong wrote: > > > > > > > > > > > > On Isn, 2017-11-06 at 11:56 +0100, Marek Vasut wrote: > > > > > > > > > > > > > > > On 11/06/2017 05:15 AM, Chee, Tien Fong wrote: > > > > > > > > > > > > > > > > > > > > > > > > On Ahd, 2017-11-05 at 17:43 +0100, Marek Vasut wrote: > > > > > > > > > > > > > > > > > > > > > > > > > > > > On 11/02/2017 09:20 AM, Chee, Tien Fong wrote: > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > On Rab, 2017-11-01 at 10:26 +0100, Marek Vasut wrote: > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > On 11/01/2017 10:18 AM, tien.fong.c...@intel.com > > > > > > > > > wrote: > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > From: Tien Fong Chee > > > > > > > > > > > > > > > > > > > > Generic firmware loader framework contains some > > > > > > > > > > common > > > > > > > > > > functionality > > > > > > > > > > which is factored out from splash loader. It is > > > > > > > > > > reusable by > > > > > > > > > > any > > > > > > > > > > specific driver file system firmware loader. > > > > > > > > > > Specific > > > > > > > > > > driver > > > > > > > > > > file > > > > > > > > > > system > > > > > > > > > > firmware loader handling can be defined with both > > > > > > > > > > weak > > > > > > > > > > function > > > > > > > > > > fsloader_preprocess and fs_loading. > > > > > > > > > > > > > > > > > > > > Signed-off-by: Tien Fong Chee > > > > > > > > > .com > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > --- > > > > > > > > > > common/Makefile | 1 + > > > > > > > > > > common/load_fs.c | 217 > > > > > > > > > > +++ > > > > > > > > > > +++ > > > > > > > > > > include/load_fs.h | 38 ++ > > > > > > > > > > 3 files changed, 256 insertions(+) > > > > > > > > > > create mode 100644 common/load_fs.c > > > > > > > > > > create mode 100644 include/load_fs.h > > > > > > > > > [...] > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > +int flash_select_fs_dev(struct flash_location > > > > > > > > > > *location) > > > > > > > > > Why does everything have flash_ prefix ? > > > > > > > > > > > > > > > > > I can remove the flash_ prefix, this generic FS loader > > > > > > > > should > > > > > > > > support > > > > > > > > for all filesystem instead of flash. > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > I also mentioned the API should copy the linux > > > > > > > > > firmware > > > > > > > > > loader > > > > > > > > > API. > > > > > > > > > > > > > > > > > If i'm not mistaken, you are referring firmware loader > > > > > > > > API > > > > > > > > in > > > > > > > > this > > > > > > > > link https://github.com/torvalds/linux/blob/f007cad159e > > > > > > > > 99fa > > > > > > > > 2acd > > > > > > > > 3b2e > > > > > > > > 9364 > > > > > > > > fbb32ad28b971/drivers/base/firmware_class.c#L1264. > > > > > > > > > > > > > > I would like to confirm with you whether we are talking to > > > > > > the > > > > > > same > > > > > > API > > > > > > above? > > > > > https://www.kernel.org/doc/html/v4.13/driver-api/firmware/ind > > > > > ex.h > > > > > tml > > > > > > > > > > first link on google btw . You might be able to avoid the > > > > > firmware > > > > > structure. > > > > > > > > > After assessment, i found that Linux loader is not suitable for > > > > fpga > > > > loader as fpga loader need > > > > 1) Able to program FPGA image in SPL chunk bu chunk with small > > > > memory > > > > allocatted. > > > > 2) Name of FPGA image defined in DTS, and path of FPGA image in > > > > FAT > > > > and > > > > UBI partition. > > > > > > > > Linux loader is strongly designed based on Linux environment, > > > > always > > > > assume having RFF, env support(which SPL don't have), sysfs and > > > > udev > > > > feature. > > > Sigh, you can just have some additional function call to fetch > > > smaller > > > chunks from a file, I don't think it's that hard of a problem ... > > > > > We already have that function to support smaller chunks, and it > > also > > work for single large image when enough memory is available in ver > > 1 > > series of fpga loadfs. > > > > Since the Linux loader API is totally not suitable for fpga loadfs, > > and > > also nothing i can leverage from there for fpga loadfs, could you > > please consider to accept implementation for this series patches or > > implementation for fpga loadfs series ver1? > You mean going back to completely custom non-generic altera-only ad- > hoc > interface ? I'd l
Re: [U-Boot] [PATCH 1/2] common: Generic file system firmware loader
On 11/09/2017 07:04 AM, Chee, Tien Fong wrote: > On Sel, 2017-11-07 at 10:34 +0100, Marek Vasut wrote: >> On 11/07/2017 10:03 AM, Chee, Tien Fong wrote: >>> >>> On Isn, 2017-11-06 at 11:56 +0100, Marek Vasut wrote: On 11/06/2017 05:15 AM, Chee, Tien Fong wrote: > > > On Ahd, 2017-11-05 at 17:43 +0100, Marek Vasut wrote: >> >> >> On 11/02/2017 09:20 AM, Chee, Tien Fong wrote: >>> >>> >>> >>> On Rab, 2017-11-01 at 10:26 +0100, Marek Vasut wrote: On 11/01/2017 10:18 AM, tien.fong.c...@intel.com wrote: > > > > > From: Tien Fong Chee > > Generic firmware loader framework contains some common > functionality > which is factored out from splash loader. It is > reusable by > any > specific driver file system firmware loader. Specific > driver > file > system > firmware loader handling can be defined with both weak > function > fsloader_preprocess and fs_loading. > > Signed-off-by: Tien Fong Chee > > --- > common/Makefile | 1 + > common/load_fs.c | 217 > ++ > include/load_fs.h | 38 ++ > 3 files changed, 256 insertions(+) > create mode 100644 common/load_fs.c > create mode 100644 include/load_fs.h [...] > > > > > +int flash_select_fs_dev(struct flash_location > *location) Why does everything have flash_ prefix ? >>> I can remove the flash_ prefix, this generic FS loader >>> should >>> support >>> for all filesystem instead of flash. >>> I also mentioned the API should copy the linux firmware loader API. >>> If i'm not mistaken, you are referring firmware loader API >>> in >>> this >>> link https://github.com/torvalds/linux/blob/f007cad159e99fa >>> 2acd >>> 3b2e >>> 9364 >>> fbb32ad28b971/drivers/base/firmware_class.c#L1264. >>> > I would like to confirm with you whether we are talking to the > same > API > above? https://www.kernel.org/doc/html/v4.13/driver-api/firmware/index.h tml first link on google btw . You might be able to avoid the firmware structure. >>> After assessment, i found that Linux loader is not suitable for >>> fpga >>> loader as fpga loader need >>> 1) Able to program FPGA image in SPL chunk bu chunk with small >>> memory >>> allocatted. >>> 2) Name of FPGA image defined in DTS, and path of FPGA image in FAT >>> and >>> UBI partition. >>> >>> Linux loader is strongly designed based on Linux environment, >>> always >>> assume having RFF, env support(which SPL don't have), sysfs and >>> udev >>> feature. >> Sigh, you can just have some additional function call to fetch >> smaller >> chunks from a file, I don't think it's that hard of a problem ... >> > We already have that function to support smaller chunks, and it also > work for single large image when enough memory is available in ver 1 > series of fpga loadfs. > > Since the Linux loader API is totally not suitable for fpga loadfs, and > also nothing i can leverage from there for fpga loadfs, could you > please consider to accept implementation for this series patches or > implementation for fpga loadfs series ver1? You mean going back to completely custom non-generic altera-only ad-hoc interface ? I'd like to hear opinion of the others on CC ... -- Best regards, Marek Vasut ___ U-Boot mailing list U-Boot@lists.denx.de https://lists.denx.de/listinfo/u-boot
Re: [U-Boot] [PATCH 1/2] common: Generic file system firmware loader
On Sel, 2017-11-07 at 10:34 +0100, Marek Vasut wrote: > On 11/07/2017 10:03 AM, Chee, Tien Fong wrote: > > > > On Isn, 2017-11-06 at 11:56 +0100, Marek Vasut wrote: > > > > > > On 11/06/2017 05:15 AM, Chee, Tien Fong wrote: > > > > > > > > > > > > On Ahd, 2017-11-05 at 17:43 +0100, Marek Vasut wrote: > > > > > > > > > > > > > > > On 11/02/2017 09:20 AM, Chee, Tien Fong wrote: > > > > > > > > > > > > > > > > > > > > > > > > On Rab, 2017-11-01 at 10:26 +0100, Marek Vasut wrote: > > > > > > > > > > > > > > > > > > > > > > > > > > > > On 11/01/2017 10:18 AM, tien.fong.c...@intel.com wrote: > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > From: Tien Fong Chee > > > > > > > > > > > > > > > > Generic firmware loader framework contains some common > > > > > > > > functionality > > > > > > > > which is factored out from splash loader. It is > > > > > > > > reusable by > > > > > > > > any > > > > > > > > specific driver file system firmware loader. Specific > > > > > > > > driver > > > > > > > > file > > > > > > > > system > > > > > > > > firmware loader handling can be defined with both weak > > > > > > > > function > > > > > > > > fsloader_preprocess and fs_loading. > > > > > > > > > > > > > > > > Signed-off-by: Tien Fong Chee > > > > > > > > > > > > > > > > --- > > > > > > > > common/Makefile | 1 + > > > > > > > > common/load_fs.c | 217 > > > > > > > > ++ > > > > > > > > include/load_fs.h | 38 ++ > > > > > > > > 3 files changed, 256 insertions(+) > > > > > > > > create mode 100644 common/load_fs.c > > > > > > > > create mode 100644 include/load_fs.h > > > > > > > [...] > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > +int flash_select_fs_dev(struct flash_location > > > > > > > > *location) > > > > > > > Why does everything have flash_ prefix ? > > > > > > > > > > > > > I can remove the flash_ prefix, this generic FS loader > > > > > > should > > > > > > support > > > > > > for all filesystem instead of flash. > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > I also mentioned the API should copy the linux firmware > > > > > > > loader > > > > > > > API. > > > > > > > > > > > > > If i'm not mistaken, you are referring firmware loader API > > > > > > in > > > > > > this > > > > > > link https://github.com/torvalds/linux/blob/f007cad159e99fa > > > > > > 2acd > > > > > > 3b2e > > > > > > 9364 > > > > > > fbb32ad28b971/drivers/base/firmware_class.c#L1264. > > > > > > > > > > I would like to confirm with you whether we are talking to the > > > > same > > > > API > > > > above? > > > https://www.kernel.org/doc/html/v4.13/driver-api/firmware/index.h > > > tml > > > > > > first link on google btw . You might be able to avoid the > > > firmware > > > structure. > > > > > After assessment, i found that Linux loader is not suitable for > > fpga > > loader as fpga loader need > > 1) Able to program FPGA image in SPL chunk bu chunk with small > > memory > > allocatted. > > 2) Name of FPGA image defined in DTS, and path of FPGA image in FAT > > and > > UBI partition. > > > > Linux loader is strongly designed based on Linux environment, > > always > > assume having RFF, env support(which SPL don't have), sysfs and > > udev > > feature. > Sigh, you can just have some additional function call to fetch > smaller > chunks from a file, I don't think it's that hard of a problem ... > We already have that function to support smaller chunks, and it also work for single large image when enough memory is available in ver 1 series of fpga loadfs. Since the Linux loader API is totally not suitable for fpga loadfs, and also nothing i can leverage from there for fpga loadfs, could you please consider to accept implementation for this series patches or implementation for fpga loadfs series ver1? Thanks. ___ U-Boot mailing list U-Boot@lists.denx.de https://lists.denx.de/listinfo/u-boot
Re: [U-Boot] [PATCH 1/2] common: Generic file system firmware loader
On 11/07/2017 10:03 AM, Chee, Tien Fong wrote: > On Isn, 2017-11-06 at 11:56 +0100, Marek Vasut wrote: >> On 11/06/2017 05:15 AM, Chee, Tien Fong wrote: >>> >>> On Ahd, 2017-11-05 at 17:43 +0100, Marek Vasut wrote: On 11/02/2017 09:20 AM, Chee, Tien Fong wrote: > > > On Rab, 2017-11-01 at 10:26 +0100, Marek Vasut wrote: >> >> >> On 11/01/2017 10:18 AM, tien.fong.c...@intel.com wrote: >>> >>> >>> >>> From: Tien Fong Chee >>> >>> Generic firmware loader framework contains some common >>> functionality >>> which is factored out from splash loader. It is reusable by >>> any >>> specific driver file system firmware loader. Specific >>> driver >>> file >>> system >>> firmware loader handling can be defined with both weak >>> function >>> fsloader_preprocess and fs_loading. >>> >>> Signed-off-by: Tien Fong Chee >>> --- >>> common/Makefile | 1 + >>> common/load_fs.c | 217 >>> ++ >>> include/load_fs.h | 38 ++ >>> 3 files changed, 256 insertions(+) >>> create mode 100644 common/load_fs.c >>> create mode 100644 include/load_fs.h >> [...] >> >>> >>> >>> >>> +int flash_select_fs_dev(struct flash_location *location) >> Why does everything have flash_ prefix ? >> > I can remove the flash_ prefix, this generic FS loader should > support > for all filesystem instead of flash. > >> >> >> I also mentioned the API should copy the linux firmware >> loader >> API. >> > If i'm not mistaken, you are referring firmware loader API in > this > link https://github.com/torvalds/linux/blob/f007cad159e99fa2acd > 3b2e > 9364 > fbb32ad28b971/drivers/base/firmware_class.c#L1264. > >>> I would like to confirm with you whether we are talking to the same >>> API >>> above? >> https://www.kernel.org/doc/html/v4.13/driver-api/firmware/index.html >> >> first link on google btw . You might be able to avoid the firmware >> structure. >> > After assessment, i found that Linux loader is not suitable for fpga > loader as fpga loader need > 1) Able to program FPGA image in SPL chunk bu chunk with small memory > allocatted. > 2) Name of FPGA image defined in DTS, and path of FPGA image in FAT and > UBI partition. > > Linux loader is strongly designed based on Linux environment, always > assume having RFF, env support(which SPL don't have), sysfs and udev > feature. Sigh, you can just have some additional function call to fetch smaller chunks from a file, I don't think it's that hard of a problem ... -- Best regards, Marek Vasut ___ U-Boot mailing list U-Boot@lists.denx.de https://lists.denx.de/listinfo/u-boot
Re: [U-Boot] [PATCH 1/2] common: Generic file system firmware loader
On Isn, 2017-11-06 at 11:56 +0100, Marek Vasut wrote: > On 11/06/2017 05:15 AM, Chee, Tien Fong wrote: > > > > On Ahd, 2017-11-05 at 17:43 +0100, Marek Vasut wrote: > > > > > > On 11/02/2017 09:20 AM, Chee, Tien Fong wrote: > > > > > > > > > > > > On Rab, 2017-11-01 at 10:26 +0100, Marek Vasut wrote: > > > > > > > > > > > > > > > On 11/01/2017 10:18 AM, tien.fong.c...@intel.com wrote: > > > > > > > > > > > > > > > > > > > > > > > > From: Tien Fong Chee > > > > > > > > > > > > Generic firmware loader framework contains some common > > > > > > functionality > > > > > > which is factored out from splash loader. It is reusable by > > > > > > any > > > > > > specific driver file system firmware loader. Specific > > > > > > driver > > > > > > file > > > > > > system > > > > > > firmware loader handling can be defined with both weak > > > > > > function > > > > > > fsloader_preprocess and fs_loading. > > > > > > > > > > > > Signed-off-by: Tien Fong Chee > > > > > > --- > > > > > > common/Makefile | 1 + > > > > > > common/load_fs.c | 217 > > > > > > ++ > > > > > > include/load_fs.h | 38 ++ > > > > > > 3 files changed, 256 insertions(+) > > > > > > create mode 100644 common/load_fs.c > > > > > > create mode 100644 include/load_fs.h > > > > > [...] > > > > > > > > > > > > > > > > > > > > > > > > > > > > > +int flash_select_fs_dev(struct flash_location *location) > > > > > Why does everything have flash_ prefix ? > > > > > > > > > I can remove the flash_ prefix, this generic FS loader should > > > > support > > > > for all filesystem instead of flash. > > > > > > > > > > > > > > > > > > > I also mentioned the API should copy the linux firmware > > > > > loader > > > > > API. > > > > > > > > > If i'm not mistaken, you are referring firmware loader API in > > > > this > > > > link https://github.com/torvalds/linux/blob/f007cad159e99fa2acd > > > > 3b2e > > > > 9364 > > > > fbb32ad28b971/drivers/base/firmware_class.c#L1264. > > > > > > I would like to confirm with you whether we are talking to the same > > API > > above? > https://www.kernel.org/doc/html/v4.13/driver-api/firmware/index.html > > first link on google btw . You might be able to avoid the firmware > structure. > After assessment, i found that Linux loader is not suitable for fpga loader as fpga loader need 1) Able to program FPGA image in SPL chunk bu chunk with small memory allocatted. 2) Name of FPGA image defined in DTS, and path of FPGA image in FAT and UBI partition. Linux loader is strongly designed based on Linux environment, always assume having RFF, env support(which SPL don't have), sysfs and udev feature. Thanks. > [...] > ___ U-Boot mailing list U-Boot@lists.denx.de https://lists.denx.de/listinfo/u-boot
Re: [U-Boot] [PATCH 1/2] common: Generic file system firmware loader
On 11/06/2017 05:15 AM, Chee, Tien Fong wrote: > On Ahd, 2017-11-05 at 17:43 +0100, Marek Vasut wrote: >> On 11/02/2017 09:20 AM, Chee, Tien Fong wrote: >>> >>> On Rab, 2017-11-01 at 10:26 +0100, Marek Vasut wrote: On 11/01/2017 10:18 AM, tien.fong.c...@intel.com wrote: > > > From: Tien Fong Chee > > Generic firmware loader framework contains some common > functionality > which is factored out from splash loader. It is reusable by any > specific driver file system firmware loader. Specific driver > file > system > firmware loader handling can be defined with both weak function > fsloader_preprocess and fs_loading. > > Signed-off-by: Tien Fong Chee > --- > common/Makefile | 1 + > common/load_fs.c | 217 > ++ > include/load_fs.h | 38 ++ > 3 files changed, 256 insertions(+) > create mode 100644 common/load_fs.c > create mode 100644 include/load_fs.h [...] > > > +int flash_select_fs_dev(struct flash_location *location) Why does everything have flash_ prefix ? >>> I can remove the flash_ prefix, this generic FS loader should >>> support >>> for all filesystem instead of flash. >>> I also mentioned the API should copy the linux firmware loader API. >>> If i'm not mistaken, you are referring firmware loader API in this >>> link https://github.com/torvalds/linux/blob/f007cad159e99fa2acd3b2e >>> 9364 >>> fbb32ad28b971/drivers/base/firmware_class.c#L1264. >>> > I would like to confirm with you whether we are talking to the same API > above? https://www.kernel.org/doc/html/v4.13/driver-api/firmware/index.html first link on google btw . You might be able to avoid the firmware structure. [...] -- Best regards, Marek Vasut ___ U-Boot mailing list U-Boot@lists.denx.de https://lists.denx.de/listinfo/u-boot
Re: [U-Boot] [PATCH 1/2] common: Generic file system firmware loader
On Ahd, 2017-11-05 at 17:43 +0100, Marek Vasut wrote:
> On 11/02/2017 09:20 AM, Chee, Tien Fong wrote:
> >
> > On Rab, 2017-11-01 at 10:26 +0100, Marek Vasut wrote:
> > >
> > > On 11/01/2017 10:18 AM, tien.fong.c...@intel.com wrote:
> > > >
> > > >
> > > > From: Tien Fong Chee
> > > >
> > > > Generic firmware loader framework contains some common
> > > > functionality
> > > > which is factored out from splash loader. It is reusable by any
> > > > specific driver file system firmware loader. Specific driver
> > > > file
> > > > system
> > > > firmware loader handling can be defined with both weak function
> > > > fsloader_preprocess and fs_loading.
> > > >
> > > > Signed-off-by: Tien Fong Chee
> > > > ---
> > > > common/Makefile | 1 +
> > > > common/load_fs.c | 217
> > > > ++
> > > > include/load_fs.h | 38 ++
> > > > 3 files changed, 256 insertions(+)
> > > > create mode 100644 common/load_fs.c
> > > > create mode 100644 include/load_fs.h
> > > [...]
> > >
> > > >
> > > >
> > > > +int flash_select_fs_dev(struct flash_location *location)
> > > Why does everything have flash_ prefix ?
> > >
> > I can remove the flash_ prefix, this generic FS loader should
> > support
> > for all filesystem instead of flash.
> >
> > >
> > > I also mentioned the API should copy the linux firmware loader
> > > API.
> > >
> > If i'm not mistaken, you are referring firmware loader API in this
> > link https://github.com/torvalds/linux/blob/f007cad159e99fa2acd3b2e
> > 9364
> > fbb32ad28b971/drivers/base/firmware_class.c#L1264.
> >
I would like to confirm with you whether we are talking to the same API
above?
> > Actually we have almost same framework in filesystem loader
> > portion,
> > just different implementation, and Linux firmware loader is more
> > specific to Linux environment such as hard code path searching in
> > RFS.
> > The generic FS loader in this patch is much more flexible, let user
> > to
> > define their own prefer implementation.
> > Linux FS firmware loader <---> U-Boot FS firmware loader
> > -- ---
> > 1) request_firmware flash_load_fs
> > 2) _request_firmware_prepare weak fsloader_preprocess
> > 3) fw_get_filesystem_firmware weak fs_loading
> >
> The API should be the same or very similar to make porting of drivers
> from Linux easy and allow people to know only one API, not two.
>
> >
> > >
> > > >
> > > > + int res;
> > > > +
> > > > + switch (location->storage) {
> > > > + case FLASH_STORAGE_MMC:
> > > > + res = fs_set_blk_dev("mmc", location->devpart,
> > > > FS_TYPE_ANY);
> > > > + break;
> > > > + case FLASH_STORAGE_USB:
> > > > + res = fs_set_blk_dev("usb", location->devpart,
> > > > FS_TYPE_ANY);
> > > > + break;
> > > > + case FLASH_STORAGE_SATA:
> > > > + res = fs_set_blk_dev("sata", location-
> > > > >devpart,
> > > > FS_TYPE_ANY);
> > > > + break;
> > > > + case FLASH_STORAGE_NAND:
> > > > + if (location->ubivol != NULL)
> > > > + res = fs_set_blk_dev("ubi", NULL,
> > > > FS_TYPE_UBIFS);
> > > > + else
> > > > + res = -ENODEV;
> > > > + break;
> > > > + default:
> > > > + error("Error: unsupported location
> > > > storage.\n");
> > > > + return -ENODEV;
> > > > + }
> > > > +
> > > > + if (res)
> > > > + error("Error: could not access storage.\n");
> > > > +
> > > > + return res;
> > > > +}
> > > > +
> > > > +#ifndef CONFIG_SPL_BUILD
> > > > +#ifdef CONFIG_USB_STORAGE
> > > This looks wrong, the USB can be supported in SPL no problem. And
> > > this
> > Technically, USB can be supported in SPL, but the build for USB in
> > SPL
> > is not supported yet.
> > >
> > > USB init shouldn't be duplicated here IMO.
> > >
> > This is just for the case USB init is not yet started, but loader
> > is
> > called 1st.
> I am not asking WHY this is needed. I suspect we have this code
> somewhere already, so it's a duplicate here.
>
___
U-Boot mailing list
U-Boot@lists.denx.de
https://lists.denx.de/listinfo/u-boot
Re: [U-Boot] [PATCH 1/2] common: Generic file system firmware loader
On 11/02/2017 09:20 AM, Chee, Tien Fong wrote:
> On Rab, 2017-11-01 at 10:26 +0100, Marek Vasut wrote:
>> On 11/01/2017 10:18 AM, tien.fong.c...@intel.com wrote:
>>>
>>> From: Tien Fong Chee
>>>
>>> Generic firmware loader framework contains some common
>>> functionality
>>> which is factored out from splash loader. It is reusable by any
>>> specific driver file system firmware loader. Specific driver file
>>> system
>>> firmware loader handling can be defined with both weak function
>>> fsloader_preprocess and fs_loading.
>>>
>>> Signed-off-by: Tien Fong Chee
>>> ---
>>> common/Makefile | 1 +
>>> common/load_fs.c | 217
>>> ++
>>> include/load_fs.h | 38 ++
>>> 3 files changed, 256 insertions(+)
>>> create mode 100644 common/load_fs.c
>>> create mode 100644 include/load_fs.h
>> [...]
>>
>>>
>>> +int flash_select_fs_dev(struct flash_location *location)
>> Why does everything have flash_ prefix ?
>>
> I can remove the flash_ prefix, this generic FS loader should support
> for all filesystem instead of flash.
>
>> I also mentioned the API should copy the linux firmware loader API.
>>
> If i'm not mistaken, you are referring firmware loader API in this
> link https://github.com/torvalds/linux/blob/f007cad159e99fa2acd3b2e9364
> fbb32ad28b971/drivers/base/firmware_class.c#L1264.
>
> Actually we have almost same framework in filesystem loader portion,
> just different implementation, and Linux firmware loader is more
> specific to Linux environment such as hard code path searching in RFS.
> The generic FS loader in this patch is much more flexible, let user to
> define their own prefer implementation.
> Linux FS firmware loader <---> U-Boot FS firmware loader
> -- ---
> 1) request_firmware flash_load_fs
> 2) _request_firmware_prepare weak fsloader_preprocess
> 3) fw_get_filesystem_firmware weak fs_loading
The API should be the same or very similar to make porting of drivers
from Linux easy and allow people to know only one API, not two.
>>> + int res;
>>> +
>>> + switch (location->storage) {
>>> + case FLASH_STORAGE_MMC:
>>> + res = fs_set_blk_dev("mmc", location->devpart,
>>> FS_TYPE_ANY);
>>> + break;
>>> + case FLASH_STORAGE_USB:
>>> + res = fs_set_blk_dev("usb", location->devpart,
>>> FS_TYPE_ANY);
>>> + break;
>>> + case FLASH_STORAGE_SATA:
>>> + res = fs_set_blk_dev("sata", location->devpart,
>>> FS_TYPE_ANY);
>>> + break;
>>> + case FLASH_STORAGE_NAND:
>>> + if (location->ubivol != NULL)
>>> + res = fs_set_blk_dev("ubi", NULL,
>>> FS_TYPE_UBIFS);
>>> + else
>>> + res = -ENODEV;
>>> + break;
>>> + default:
>>> + error("Error: unsupported location storage.\n");
>>> + return -ENODEV;
>>> + }
>>> +
>>> + if (res)
>>> + error("Error: could not access storage.\n");
>>> +
>>> + return res;
>>> +}
>>> +
>>> +#ifndef CONFIG_SPL_BUILD
>>> +#ifdef CONFIG_USB_STORAGE
>> This looks wrong, the USB can be supported in SPL no problem. And
>> this
> Technically, USB can be supported in SPL, but the build for USB in SPL
> is not supported yet.
>> USB init shouldn't be duplicated here IMO.
>>
> This is just for the case USB init is not yet started, but loader is
> called 1st.
I am not asking WHY this is needed. I suspect we have this code
somewhere already, so it's a duplicate here.
--
Best regards,
Marek Vasut
___
U-Boot mailing list
U-Boot@lists.denx.de
https://lists.denx.de/listinfo/u-boot
Re: [U-Boot] [PATCH 1/2] common: Generic file system firmware loader
On Rab, 2017-11-01 at 10:26 +0100, Marek Vasut wrote:
> On 11/01/2017 10:18 AM, tien.fong.c...@intel.com wrote:
> >
> > From: Tien Fong Chee
> >
> > Generic firmware loader framework contains some common
> > functionality
> > which is factored out from splash loader. It is reusable by any
> > specific driver file system firmware loader. Specific driver file
> > system
> > firmware loader handling can be defined with both weak function
> > fsloader_preprocess and fs_loading.
> >
> > Signed-off-by: Tien Fong Chee
> > ---
> > common/Makefile | 1 +
> > common/load_fs.c | 217
> > ++
> > include/load_fs.h | 38 ++
> > 3 files changed, 256 insertions(+)
> > create mode 100644 common/load_fs.c
> > create mode 100644 include/load_fs.h
> [...]
>
> >
> > +int flash_select_fs_dev(struct flash_location *location)
> Why does everything have flash_ prefix ?
>
I can remove the flash_ prefix, this generic FS loader should support
for all filesystem instead of flash.
> I also mentioned the API should copy the linux firmware loader API.
>
If i'm not mistaken, you are referring firmware loader API in this
link https://github.com/torvalds/linux/blob/f007cad159e99fa2acd3b2e9364
fbb32ad28b971/drivers/base/firmware_class.c#L1264.
Actually we have almost same framework in filesystem loader portion,
just different implementation, and Linux firmware loader is more
specific to Linux environment such as hard code path searching in RFS.
The generic FS loader in this patch is much more flexible, let user to
define their own prefer implementation.
Linux FS firmware loader <---> U-Boot FS firmware loader
-- ---
1) request_firmware flash_load_fs
2) _request_firmware_prepare weak fsloader_preprocess
3) fw_get_filesystem_firmware weak fs_loading
> > + int res;
> > +
> > + switch (location->storage) {
> > + case FLASH_STORAGE_MMC:
> > + res = fs_set_blk_dev("mmc", location->devpart,
> > FS_TYPE_ANY);
> > + break;
> > + case FLASH_STORAGE_USB:
> > + res = fs_set_blk_dev("usb", location->devpart,
> > FS_TYPE_ANY);
> > + break;
> > + case FLASH_STORAGE_SATA:
> > + res = fs_set_blk_dev("sata", location->devpart,
> > FS_TYPE_ANY);
> > + break;
> > + case FLASH_STORAGE_NAND:
> > + if (location->ubivol != NULL)
> > + res = fs_set_blk_dev("ubi", NULL,
> > FS_TYPE_UBIFS);
> > + else
> > + res = -ENODEV;
> > + break;
> > + default:
> > + error("Error: unsupported location storage.\n");
> > + return -ENODEV;
> > + }
> > +
> > + if (res)
> > + error("Error: could not access storage.\n");
> > +
> > + return res;
> > +}
> > +
> > +#ifndef CONFIG_SPL_BUILD
> > +#ifdef CONFIG_USB_STORAGE
> This looks wrong, the USB can be supported in SPL no problem. And
> this
Technically, USB can be supported in SPL, but the build for USB in SPL
is not supported yet.
> USB init shouldn't be duplicated here IMO.
>
This is just for the case USB init is not yet started, but loader is
called 1st.
> >
> > +static int flash_init_usb(void)
> > +{
> > + int err;
> > +
> > + err = usb_init();
> > + if (err)
> > + return err;
> > +
> > +#ifndef CONFIG_DM_USB
> > + err = usb_stor_scan(1) < 0 ? -ENODEV : 0;
> > +#endif
> > +
> > + return err;
> > +}
> > +#else
> > +static inline int flash_init_usb(void)
> Don't use inline , the compiler can decide better.
>
Okay.
> [...]
>
>
___
U-Boot mailing list
U-Boot@lists.denx.de
https://lists.denx.de/listinfo/u-boot
Re: [U-Boot] [PATCH 1/2] common: Generic file system firmware loader
On 11/01/2017 10:18 AM, tien.fong.c...@intel.com wrote:
> From: Tien Fong Chee
>
> Generic firmware loader framework contains some common functionality
> which is factored out from splash loader. It is reusable by any
> specific driver file system firmware loader. Specific driver file system
> firmware loader handling can be defined with both weak function
> fsloader_preprocess and fs_loading.
>
> Signed-off-by: Tien Fong Chee
> ---
> common/Makefile | 1 +
> common/load_fs.c | 217
> ++
> include/load_fs.h | 38 ++
> 3 files changed, 256 insertions(+)
> create mode 100644 common/load_fs.c
> create mode 100644 include/load_fs.h
[...]
> +int flash_select_fs_dev(struct flash_location *location)
Why does everything have flash_ prefix ?
I also mentioned the API should copy the linux firmware loader API.
> +{
> + int res;
> +
> + switch (location->storage) {
> + case FLASH_STORAGE_MMC:
> + res = fs_set_blk_dev("mmc", location->devpart, FS_TYPE_ANY);
> + break;
> + case FLASH_STORAGE_USB:
> + res = fs_set_blk_dev("usb", location->devpart, FS_TYPE_ANY);
> + break;
> + case FLASH_STORAGE_SATA:
> + res = fs_set_blk_dev("sata", location->devpart, FS_TYPE_ANY);
> + break;
> + case FLASH_STORAGE_NAND:
> + if (location->ubivol != NULL)
> + res = fs_set_blk_dev("ubi", NULL, FS_TYPE_UBIFS);
> + else
> + res = -ENODEV;
> + break;
> + default:
> + error("Error: unsupported location storage.\n");
> + return -ENODEV;
> + }
> +
> + if (res)
> + error("Error: could not access storage.\n");
> +
> + return res;
> +}
> +
> +#ifndef CONFIG_SPL_BUILD
> +#ifdef CONFIG_USB_STORAGE
This looks wrong, the USB can be supported in SPL no problem. And this
USB init shouldn't be duplicated here IMO.
> +static int flash_init_usb(void)
> +{
> + int err;
> +
> + err = usb_init();
> + if (err)
> + return err;
> +
> +#ifndef CONFIG_DM_USB
> + err = usb_stor_scan(1) < 0 ? -ENODEV : 0;
> +#endif
> +
> + return err;
> +}
> +#else
> +static inline int flash_init_usb(void)
Don't use inline , the compiler can decide better.
[...]
--
Best regards,
Marek Vasut
___
U-Boot mailing list
U-Boot@lists.denx.de
https://lists.denx.de/listinfo/u-boot
[U-Boot] [PATCH 1/2] common: Generic file system firmware loader
From: Tien Fong Chee
Generic firmware loader framework contains some common functionality
which is factored out from splash loader. It is reusable by any
specific driver file system firmware loader. Specific driver file system
firmware loader handling can be defined with both weak function
fsloader_preprocess and fs_loading.
Signed-off-by: Tien Fong Chee
---
common/Makefile | 1 +
common/load_fs.c | 217 ++
include/load_fs.h | 38 ++
3 files changed, 256 insertions(+)
create mode 100644 common/load_fs.c
create mode 100644 include/load_fs.h
diff --git a/common/Makefile b/common/Makefile
index 801ea31..89f9365 100644
--- a/common/Makefile
+++ b/common/Makefile
@@ -130,3 +130,4 @@ obj-$(CONFIG_CMD_DFU) += dfu.o
obj-y += command.o
obj-y += s_record.o
obj-y += xyzModem.o
+obj-y += load_fs.o
diff --git a/common/load_fs.c b/common/load_fs.c
new file mode 100644
index 000..112b4f6
--- /dev/null
+++ b/common/load_fs.c
@@ -0,0 +1,217 @@
+/*
+ * Copyright (C) 2017 Intel Corporation
+ *
+ * SPDX-License-Identifier:GPL-2.0
+ */
+
+#include
+#include
+#include
+#include
+#include
+#include
+#include
+#include
+#include
+
+int flash_select_fs_dev(struct flash_location *location)
+{
+ int res;
+
+ switch (location->storage) {
+ case FLASH_STORAGE_MMC:
+ res = fs_set_blk_dev("mmc", location->devpart, FS_TYPE_ANY);
+ break;
+ case FLASH_STORAGE_USB:
+ res = fs_set_blk_dev("usb", location->devpart, FS_TYPE_ANY);
+ break;
+ case FLASH_STORAGE_SATA:
+ res = fs_set_blk_dev("sata", location->devpart, FS_TYPE_ANY);
+ break;
+ case FLASH_STORAGE_NAND:
+ if (location->ubivol != NULL)
+ res = fs_set_blk_dev("ubi", NULL, FS_TYPE_UBIFS);
+ else
+ res = -ENODEV;
+ break;
+ default:
+ error("Error: unsupported location storage.\n");
+ return -ENODEV;
+ }
+
+ if (res)
+ error("Error: could not access storage.\n");
+
+ return res;
+}
+
+#ifndef CONFIG_SPL_BUILD
+#ifdef CONFIG_USB_STORAGE
+static int flash_init_usb(void)
+{
+ int err;
+
+ err = usb_init();
+ if (err)
+ return err;
+
+#ifndef CONFIG_DM_USB
+ err = usb_stor_scan(1) < 0 ? -ENODEV : 0;
+#endif
+
+ return err;
+}
+#else
+static inline int flash_init_usb(void)
+{
+ error("Error: Cannot load flash image: no USB support\n");
+ return -ENOSYS;
+}
+#endif
+#endif
+
+#ifdef CONFIG_SATA
+static int flash_init_sata(void)
+{
+ return sata_probe(0);
+}
+#else
+static inline int flash_init_sata(void)
+{
+ error("Error: Cannot load flash image: no SATA support\n");
+ return -ENOSYS;
+}
+#endif
+
+#ifdef CONFIG_CMD_UBIFS
+static int flash_mount_ubifs(struct flash_location *location)
+{
+ int res;
+ char cmd[32];
+
+ sprintf(cmd, "ubi part %s", location->mtdpart);
+ res = run_command(cmd, 0);
+ if (res)
+ return res;
+
+ sprintf(cmd, "ubifsmount %s", location->ubivol);
+ res = run_command(cmd, 0);
+
+ return res;
+}
+
+static inline int flash_umount_ubifs(void)
+{
+ return run_command("ubifsumount", 0);
+}
+#else
+static inline int flash_mount_ubifs(struct flash_location *location)
+{
+ error("Error: Cannot load flash image: no UBIFS support\n");
+ return -ENOSYS;
+}
+
+static inline int flash_umount_ubifs(void)
+{
+ error("Error: Cannot unmount UBIFS: no UBIFS support\n");
+ return -ENOSYS;
+}
+#endif
+
+/**
+ * fsloader_preprocess - Any prepocessing before calling filesystem loader such
+ * getting filename, and flash partition information.
+ *
+ * @locations: An array of supported flash locations. Contains default
+ * flash setting for the file image.
+ * @file_info: Description and attributes to the image.
+ * Could be structure pointer, and any type pointer.
+ * @filename: Image filename in flashes. Storing the file image name
+ * to this pointer after retriving the name from DTB,
+ * or environment or source code.
+ * @load_addr: Target location image loaded to.
+ *
+ * @return:If 0, processing is succesfull. Filename pointer
+ * contains valid filename.
+ * If -ve, processing is failed.
+ */
+__weak int fsloader_preprocess(struct flash_location *location,
+ void *file_info, char **filename,
+ u32 load_addr)
+{
+ return 0;
+}
+
+/*
+ * fs_loading - Place for implementing whatever specific driver to
+ * the target HW such as program raw binary file to FPGA.
+ *
+ * @locations: An array of supported
