| Submitter | Simon Glass |
|---|---|
| Date | Feb. 16, 2013, 12:14 a.m. |
| Message ID | <CAPnjgZ2P6sBDXiwXW2TeCdjADMhkN5iNBGrpZbtvwMqUtYVVxA@mail.gmail.com> |
| Download | mbox |
| Permalink | /patch/220907/ |
| State | Rejected, archived |
| Delegated to: | Tom Rini |
| Headers | show |
Pull-request
git://git.denx.de/u-boot-x86.git masterComments
Dear Simon Glass, In message <CAPnjgZ2P6sBDXiwXW2TeCdjADMhkN5iNBGrpZbtvwMqUtYVVxA@mail.gmail.com> you wrote: > Hi Tom, > > This series includes the sandbox map_sysmem() feature, and gets the > memory and hashing functions running on sandbox to allow testing/code > coverage. I have run it through buildman and it seems clean, with the > proviso that I don't have fully-working toolchains for all > architectures. NAK. It is not correct to push changes that affect global code through a arch-specific custodian tree, especially if the submitter of the patche(es) is identical to the custodian of the very tree, and even more so if there have been not ANY independent Acked-by: or at least Tested-by: messages. This is NOT how the peer review process is supposed to work!! Especially as a custodian you must not do such things. Best regards, Wolfgang Denk
Hi Wolfgang, On Sun, Feb 17, 2013 at 12:58 PM, Wolfgang Denk <wd@denx.de> wrote: > Dear Simon Glass, > > In message <CAPnjgZ2P6sBDXiwXW2TeCdjADMhkN5iNBGrpZbtvwMqUtYVVxA@mail.gmail.com> you wrote: >> Hi Tom, >> >> This series includes the sandbox map_sysmem() feature, and gets the >> memory and hashing functions running on sandbox to allow testing/code >> coverage. I have run it through buildman and it seems clean, with the >> proviso that I don't have fully-working toolchains for all >> architectures. > > NAK. It is not correct to push changes that affect global code > through a arch-specific custodian tree, especially if the submitter > of the patche(es) is identical to the custodian of the very tree, and > even more so if there have been not ANY independent Acked-by: or at > least Tested-by: messages. > > This is NOT how the peer review process is supposed to work!! > > Especially as a custodian you must not do such things. OK, I was not quite sure what to do, so may have misunderstood Tom's instructions - there is a short thread here http://comments.gmane.org/gmane.comp.boot-loaders.u-boot/153342 I have created a patchwork bundle instead. http://patchwork.ozlabs.org/bundle/sjg/sandbox-mem/ Only one patch was Acked, so it could certainly use a few more eyes. However, it has been on the list for nearly two months, and I feel that applying things too close to the next release doesn't give people a lot of time to find problems. Regards, Simon > > > Best regards, > > Wolfgang Denk > > -- > DENX Software Engineering GmbH, MD: Wolfgang Denk & Detlev Zundel > HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany > Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd@denx.de > Pray to God, but keep rowing to shore. - Russian Proverb
On Sun, Feb 17, 2013 at 01:32:58PM -0800, Simon Glass wrote: > Hi Wolfgang, > > On Sun, Feb 17, 2013 at 12:58 PM, Wolfgang Denk <wd@denx.de> wrote: > > Dear Simon Glass, > > > > In message <CAPnjgZ2P6sBDXiwXW2TeCdjADMhkN5iNBGrpZbtvwMqUtYVVxA@mail.gmail.com> you wrote: > >> Hi Tom, > >> > >> This series includes the sandbox map_sysmem() feature, and gets the > >> memory and hashing functions running on sandbox to allow testing/code > >> coverage. I have run it through buildman and it seems clean, with the > >> proviso that I don't have fully-working toolchains for all > >> architectures. > > > > NAK. It is not correct to push changes that affect global code > > through a arch-specific custodian tree, especially if the submitter > > of the patche(es) is identical to the custodian of the very tree, and > > even more so if there have been not ANY independent Acked-by: or at > > least Tested-by: messages. > > > > This is NOT how the peer review process is supposed to work!! > > > > Especially as a custodian you must not do such things. > > OK, I was not quite sure what to do, so may have misunderstood Tom's > instructions - there is a short thread here > http://comments.gmane.org/gmane.comp.boot-loaders.u-boot/153342 > > I have created a patchwork bundle instead. > > http://patchwork.ozlabs.org/bundle/sjg/sandbox-mem/ OK, I thought I said, but maybe I didn't, I'm OK with re-using the tree, but _not_ the master branch, u-boot-x86/sandbox would have been fine.
On Mon, Feb 18, 2013 at 7:52 PM, Tom Rini <trini@ti.com> wrote: > OK, I thought I said, but maybe I didn't, I'm OK with re-using the tree, > but _not_ the master branch, u-boot-x86/sandbox would have been fine. Personally I'd prefer another tree as done for other custodians. It makes life of new developers easier.
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On 02/18/2013 06:30 PM, Otavio Salvador wrote: > On Mon, Feb 18, 2013 at 7:52 PM, Tom Rini <trini@ti.com> wrote: >> OK, I thought I said, but maybe I didn't, I'm OK with re-using >> the tree, but _not_ the master branch, u-boot-x86/sandbox would >> have been fine. > > Personally I'd prefer another tree as done for other custodians. > It makes life of new developers easier. It doesn't scale, however. If I had my wish and we were starting this afresh, I'd go with user repositories rather than subject repositories. Using Simon as the example, I don't think he needs one for sandbox, one for patman (and other tools) and one for x86. I'd rather pull .../sjc/for-trini/x86-whatever-vs-something. - -- Tom -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.11 (GNU/Linux) Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/ iQIcBAEBAgAGBQJRIr00AAoJENk4IS6UOR1W5BsP/0XxGgvNqvp/Od5k1JnrR9EW mqf6xRV7IZ6MXwPBAK7/WBaAerEZ79vx2KSezQxejkxzJlTYVgiltJHf9GNCM3xh aenk/RGsyjjPmvZXTY/kR79x3/tMMdEu3xHaLb9F2a62qWfOAjQAcjqWtfwN1mSP TOgnEenxYovihC8hqQA+Qo6PjRwTQJStGapUCwxWinXAD1CWxcp3QdlHr8I2T6Ib TYIBSzT5iM/9LSdexh0Z8HOqQ0Mdu91znbJZCROkSWN5E9PM/oRaiXoWfSF6zWZy mjhmI9V+Egl9SOhJU3XL6Q2Zjs98jsnQMIELczHrHFxidWjbdopYD5GOEfx59A+z R8zZc59TSe1ocWdoJOkToy33iiXhWSUJR3ig6fmofVV7IXF/i9yO07GrlpW+CUip GVxAUaZdSgPfNtIKXQ10zwzO3VGRgxk2eLs5zb+cMR/wc/gy8cqQY9J9GJwF/k7t cysCzTW+iaSBaXwYSgVIscO7e7B9x+rt7Py+1MkkYegVE5N5Z2Igh8J5Z/tHe1Fi A+GuvkcX9aytvSiBtKjqBbe0pSc10h1EfsmAfhH1F/Go94RA+58cPMNPmzN9KkBj 6+TahbMkzk5vsHcLosio6Oj0ZNS0Xo6w/XAZGyhep0gl61YSZNwUGx5pBbcQ6OPi 2I6hs4o9GUZV2id6IZU9 =dJRp -----END PGP SIGNATURE-----
On Mon, Feb 18, 2013 at 8:45 PM, Tom Rini <trini@ti.com> wrote: > -----BEGIN PGP SIGNED MESSAGE----- > Hash: SHA1 > > On 02/18/2013 06:30 PM, Otavio Salvador wrote: >> On Mon, Feb 18, 2013 at 7:52 PM, Tom Rini <trini@ti.com> wrote: >>> OK, I thought I said, but maybe I didn't, I'm OK with re-using >>> the tree, but _not_ the master branch, u-boot-x86/sandbox would >>> have been fine. >> >> Personally I'd prefer another tree as done for other custodians. >> It makes life of new developers easier. > > It doesn't scale, however. If I had my wish and we were starting this > afresh, I'd go with user repositories rather than subject > repositories. Using Simon as the example, I don't think he needs one > for sandbox, one for patman (and other tools) and one for x86. I'd > rather pull .../sjc/for-trini/x86-whatever-vs-something. The hassle to send to separated branches is the same for different remotes; what concerns me is a new developer to try to find patman or sandbox pending patches and do not realize it is at x86 tree. This is confusing.
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On 02/18/2013 06:48 PM, Otavio Salvador wrote: > On Mon, Feb 18, 2013 at 8:45 PM, Tom Rini <trini@ti.com> wrote: >> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 >> >> On 02/18/2013 06:30 PM, Otavio Salvador wrote: >>> On Mon, Feb 18, 2013 at 7:52 PM, Tom Rini <trini@ti.com> >>> wrote: >>>> OK, I thought I said, but maybe I didn't, I'm OK with >>>> re-using the tree, but _not_ the master branch, >>>> u-boot-x86/sandbox would have been fine. >>> >>> Personally I'd prefer another tree as done for other >>> custodians. It makes life of new developers easier. >> >> It doesn't scale, however. If I had my wish and we were starting >> this afresh, I'd go with user repositories rather than subject >> repositories. Using Simon as the example, I don't think he needs >> one for sandbox, one for patman (and other tools) and one for >> x86. I'd rather pull >> .../sjc/for-trini/x86-whatever-vs-something. > > The hassle to send to separated branches is the same for different > remotes; what concerns me is a new developer to try to find patman > or sandbox pending patches and do not realize it is at x86 tree. > This is confusing. It's not that great in kernel-land, I agree. But at least for U-Boot I hope we would be able to keep the number, and possibly as/more importantly, the time trees are not in master (or next) but have good things in them. - -- Tom -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.11 (GNU/Linux) Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/ iQIcBAEBAgAGBQJRIr6WAAoJENk4IS6UOR1WOOsP/2mxdE/HHOF0GMhDAo9xDPqm 24NHn5286PCGbQIizbbwBIOnb0/suFLbqNIhIXaE587z4veDFwiTR74Ns85NrDvi 2IUavB0QwpSO8dwpjykOHvo5aA8DUaM6jYxXhrgnm3fsvlZJIgrOcEHtBd3xKerq LdGrHybmjPZFhC9kK04JoICVAJb8svnWH9C+ql56QLn1/ZjwHVP3OlYv3bmx+iKG QuBtx30CmLwciJBAq6x3LlVasm76naA9S444RFPwxH0s3Eqlsl111Z9FdtAnc2HW VCgzU+GPgowayLSnMCf0RdXL5ho23vpOYsABOd+jKVCoK3VgEkSpyQXWk52vKD5h pRTqBNOl7KrVaCYcB5NC+xB/5dTUpem3qfvQ6UAbElehLDfHvI82Dd7ttPl0GMsH +aUTNdqubHtU6DmApGQDrfOyzi4u4vLSy3CX6Am/7pGpfd3h1M+gLChhLNH010H0 b9BMOWRRc8TolydImTFHuibCtEfx7HcleIujlThodTocU4aTkaUMoi3nSeJ58KcA vPJl1Y/vR8MmuS5mQr//Lzvf+nsSxQYg18crkasPqbqiiAeCAKHBkHd7ZD+ALCxY k9Plxpa/sx8DUqQUdgFxyDz5kAPCOli/BLdC6h/+7Yflp6HAwa/Ly6QoFYT6yGA3 79hxzWjtJ5/uravy4eKl =5qPb -----END PGP SIGNATURE-----
Hi Tom, On Mon, Feb 18, 2013 at 2:52 PM, Tom Rini <trini@ti.com> wrote: > On Sun, Feb 17, 2013 at 01:32:58PM -0800, Simon Glass wrote: >> Hi Wolfgang, >> >> On Sun, Feb 17, 2013 at 12:58 PM, Wolfgang Denk <wd@denx.de> wrote: >> > Dear Simon Glass, >> > >> > In message <CAPnjgZ2P6sBDXiwXW2TeCdjADMhkN5iNBGrpZbtvwMqUtYVVxA@mail.gmail.com> you wrote: >> >> Hi Tom, >> >> >> >> This series includes the sandbox map_sysmem() feature, and gets the >> >> memory and hashing functions running on sandbox to allow testing/code >> >> coverage. I have run it through buildman and it seems clean, with the >> >> proviso that I don't have fully-working toolchains for all >> >> architectures. >> > >> > NAK. It is not correct to push changes that affect global code >> > through a arch-specific custodian tree, especially if the submitter >> > of the patche(es) is identical to the custodian of the very tree, and >> > even more so if there have been not ANY independent Acked-by: or at >> > least Tested-by: messages. >> > >> > This is NOT how the peer review process is supposed to work!! >> > >> > Especially as a custodian you must not do such things. >> >> OK, I was not quite sure what to do, so may have misunderstood Tom's >> instructions - there is a short thread here >> http://comments.gmane.org/gmane.comp.boot-loaders.u-boot/153342 >> >> I have created a patchwork bundle instead. >> >> http://patchwork.ozlabs.org/bundle/sjg/sandbox-mem/ > > OK, I thought I said, but maybe I didn't, I'm OK with re-using the tree, > but _not_ the master branch, u-boot-x86/sandbox would have been fine. Yes, you said "toss it into a branch in u-boot-x86.git". It did cross my mind to use something other than master, but I wasn't sure if that was OK in U-Boot. I know for next time. Regards, Simon > > -- > Tom
Hi Tom, This series includes the sandbox map_sysmem() feature, and gets the memory and hashing functions running on sandbox to allow testing/code coverage. I have run it through buildman and it seems clean, with the proviso that I don't have fully-working toolchains for all architectures. The following changes since commit 9f024f62e4604274a23213dcee30016092e32e7b: Merge branch 'fixes' of git://git.denx.de/u-boot-mips (2013-02-15 12:23:42 -0500) are available in the git repository at: git://git.denx.de/u-boot-x86.git master for you to fetch changes up to d76bd1c2d5e75db16b239bd2d4a642673974e250: sandbox: Allow hash functions to work correctly (2013-02-15 16:06:06 -0800) ---------------------------------------------------------------- Allen Martin (1): sandbox: fix compiler warning Simon Glass (19): Tidy up error checking and fix bug in hash command Update print_buffer() to use const sandbox: Add un/map_sysmen() to deal with sandbox's ram_buf sandbox: Change memory commands to use map_physmem Split out the memory tests into separate functions Use common mtest iteration counting Fix mtest indenting Bring mtest putc() into common code Reduce casting in mtest Update set_working_fdt_addr() to use setenv_addr() common: Use new numeric setenv functions drivers: Use new numeric setenv functions net: Use new numeric setenv functions image: Use crc header file instead of C prototypes Roll crc32 into hash infrastructure sandbox: config: Enable hash functions and mtest Move CONFIG_SYS_MEMTEST_SCRATCH #ifdef to top of file sandbox: Update mtest to fix crashes sandbox: Allow hash functions to work correctly Taylor Hutt (1): sandbox: Improve sandbox serial port keyboard interface README | 9 + arch/sandbox/config.mk | 1 + arch/sandbox/cpu/os.c | 8 + arch/sandbox/cpu/start.c | 3 + arch/sandbox/include/asm/io.h | 10 + common/cmd_bootm.c | 11 +- common/cmd_cbfs.c | 4 +- common/cmd_cramfs.c | 4 +- common/cmd_fdos.c | 4 +- common/cmd_fdt.c | 11 +- common/cmd_hash.c | 4 + common/cmd_jffs2.c | 4 +- common/cmd_load.c | 12 +- common/cmd_mem.c | 798 +++++++++++++++++++++--------------------- common/cmd_mtdparts.c | 4 +- common/cmd_nand.c | 12 +- common/cmd_nvedit.c | 11 +- common/cmd_reiser.c | 4 +- common/cmd_setexpr.c | 39 ++- common/cmd_unzip.c | 4 +- common/cmd_ximg.c | 7 +- common/cmd_zfs.c | 3 +- common/cmd_zip.c | 4 +- common/hash.c | 31 +- common/image.c | 4 +- drivers/net/fm/fm.c | 4 +- drivers/serial/sandbox.c | 44 ++- fs/fs.c | 4 +- fs/ubifs/ubifs.c | 4 +- include/common.h | 29 +- include/configs/sandbox.h | 9 +- include/hash.h | 2 +- include/os.h | 10 + include/u-boot/crc.h | 11 + lib/crc32.c | 9 + lib/display_options.c | 3 +- net/net.c | 8 +- 37 files changed, 615 insertions(+), 528 deletions(-) Regards, Simon