On 9/8/22 06:11, наб wrote:
> The entire file blames back to the start of git
> (minus whitespace from the RST translation and a typo fix):
> * there are changelog comments for March 1994 through to Linux 2.5.74
> * struct tty_ldisc is two pointers nowadays, so naturally no magic
> * GDA_MAGIC is defined but unused, and it's been this way
> since start-of-git
> * M3_CARD_MAGIC isn't defined, because
> commit d56b9b9c464a10ab1ee51a4c6190a2b57b8ef7a6 ("[PATCH] The
> scheduled removal of some OSS drivers")
> removed the entire driver in 2006
> * CS_CARD_MAGIC likewise since
> commit b5d425c97f7d4e92151167b01ca038e7853c6b37 ("more scheduled OSS
> driver removal") in 2007
> * KMALLOC_MAGIC and VMALLOC_MAGIC were removed in
> commit e38e0cfa48ac38f4fe24453d2523852467c95b21 ("[ALSA] Remove
> kmalloc wrappers"), six months after start of git
> * SLAB_C_MAGIC has never even appeared in git
> (removed in in 2.4.0-test3pre6)
>
Hi,
The patch description can be improved, see below.
Nit: Use abbreviated hash when referring to commits (`git show -s
--pretty=format:"%h (\"%s\")`). You may need to set `core.abbrev=12`
in your `~/.gitconfig`.
> At a glance, the only ones still in use are in code mostly untouched
> since git (TTY, SLIP, SCSI (drivers/scsi/{arm/queue.c,ncr53c8xx.c}),
> &c., but not PPP or PTY (neither exist now)).
>
"... are in drivers which are mostly untouched..."
Also, really PPP and PTY have been burninated? Since what commit?
> A rough estimate of magic numbers in the kernel may be obtained via
> # git grep define | grep '_MAGIC\b' |
> grep -ve IOC -e ^tools/ -e '_IO[(WR]' -e _SUPER | wc -l
> 780
> there is, of course, a few other meanings of magic here,
> but a random sampling of SR_EEPROM_MAGIC, L2TP_TUNNEL_MAGIC,
> and MD_SB_MAGIC showed they are all used like magic-number.rst describes
> and are not in it
>
"A rough estimate ... is obtained by:"
"Note that the word 'magic' may have different meaning from magic
number context.
There are also magic numbers which aren't listed here, like ..."
> By all accounts, it seems magic-number.rst is a low-value historical
> relic at best and misleading cruft at worst
>
"Therefore, magic-number.rst is a low-value historial relic at best and
misleading cruft at worst, so remove the documentation."
Thanks.
--
An old man doll... just what I always wanted! - Clara
