udhcp: use poll() instead of select()

function                                             old     new   delta
udhcp_sp_read                                         65      46     -19
udhcp_sp_fd_set                                       79      54     -25
udhcpd_main                                         1530    1482     -48
udhcpc_main                                         2780    2730     -50
------------------------------------------------------------------------------
(add/remove: 0/0 grow/shrink: 0/4 up/down: 0/-142)           Total: -142 bytes

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

udhcp: do not clobber errno by signal handler

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

nc: use poll() instead of select()

function                                             old     new   delta
nc_main                                              943     866     -77

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

tls: covert i/o loop from using select() to poll()

function                                             old     new   delta
tls_run_copy_loop                                    377     282     -95

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

sendmail: allow "+" symbol in recipient. Closes 9646

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

modprobe-small: define and use DEPMOD_OPT_n (option mask)

Signed-off-by: Kang-Che Sung <explorer09@gmail.com>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

modprobe-small: document '-n' in depmod usage

Signed-off-by: Kang-Che Sung <explorer09@gmail.com>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

modprobe_small: if only MODPROBE and DEPMOD are selected, no need to test for them

function                                             old     new   delta
modprobe_main                                        321     306     -15

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

Reorder modutils config options & fix yet more dependency

- modprobe can indirectly benefit from FEATURE_2_4_MODULES and
  FEATURE_INSMOD_TRY_MAP options.
- The position of config FEATURE_INSMOD_TRY_MMAP prevented some other
  config options from indenting under FEATURE_2_4_MODULES. Reorder to
  fix this.
- FEATURE_MODPROBE_SMALL_CHECK_ALREADY_LOADED is now moved to
  Config.src under "Common options" section. (I wished to edit this
  config so that it also work with "big" modutils, but it's not done at
  the moment. Sorry.)

Signed-off-by: Kang-Che Sung <explorer09@gmail.com>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

modutils: fix config options dependency (2)

- The modprobe-small implementation of rmmod no longer chdir's to
  "/lib/modules/`uname -r`" as it was not necessary for rmmod's
  operation. (And it no longer need to die if such modules directory
  doesn't exist.)
- Configs DEFAULT_MODULES_DIR and DEFAULT_DEPMOD_FILE no longer depend
  on MODPROBE_SMALL as the latter may not enable depmod or modprobe
  that requires these configs.
- Clarify DEFAULT_DEPMOD_FILE's description regarding the ".bb" name
  suffix.

Signed-off-by: Kang-Che Sung <explorer09@gmail.com>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

httpd: use "Content-Length", not "-length"

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

Replace int -> uint to avoid signed integer overflow

An example of such an error (should be compiled with DEBUG_SANITIZE):

runtime error: left shift of 1 by 31 places cannot be represented in
type 'int'

Signed-off-by: Rostislav Skudnov <rostislav@tuxera.com>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

tls: fold AES CBC en/decryption into single functions

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

wget/tls: session_id of zero length is ok (arxiv.org responds with such)

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

ash: add INT_OFF/ON around allocations

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

ps: avoid -o stat to contain spaces. Closes 9631

function                                             old     new   delta
procps_scan                                         1227    1236      +9

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

cmdline module options can be disabled on "big" modutils

Allow module options on command line to be disabled on "big" modutils.

Config FEATURE_MODPROBE_SMALL_OPTIONS_ON_CMDLINE is renamed to
FEATURE_CMDLINE_MODULE_OPTIONS and no longer depends on !MODPROBE_SMALL

(I'm not sure if disabling this is useful on "big" modutils, but at
least the macro can serve as a marker and ensure both implementations
of same feature have consistent behavior.)

Signed-off-by: Kang-Che Sung <explorer09@gmail.com>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

modutils: remove redundant "select PLATFORM_LINUX" configs

It is enough to have only applets' configs select PLATFORM_LINUX.

Signed-off-by: Kang-Che Sung <explorer09@gmail.com>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

modprobe-small: move lsmod code out of modprobe_main()

Having lsmod code inside modprobe_main() makes some of the applet name
checking code awkward. Besides, this make busybox x86_64 binary a few
bytes smaller. :)

function                                  old     new   delta
lsmod_main                                  -      23     +23
modprobe_main                             599     564     -35
-------------------------------------------------------------------
(add/remove: 1/0 grow/shrink: 0/1 up/down: 23/-35) Total: -12 bytes

Signed-off-by: Kang-Che Sung <explorer09@gmail.com>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

modprobe-small: improve config help text

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

ash: improve / fix glob expansion

When using musl libc glob() a very long string can cause glob() to fail,
which leads to an out of memory error being raised by ash.

This can happen easily if a very long quoted string contains *, even
though no glob expansion should ever be performed on it (since it's
quoted).

Fix this by properly parsing control characters and escaping and only
accept unquoted metacharacters. While we're at it, unify this check for
libc and built-in glob expansion

Signed-off-by: Felix Fietkau <nbd@nbd.name>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

taskset: simplify code a bit; tweak --help

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

modprobe-small: fix --help texts, they are from "big" modutils

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

modutils: fix config options dependency

module.aliases and module.symbols files have no use in modprobe-small
implementation. So FEATURE_MODUTILS_ALIAS and FEATURE_MODUTILS_SYMBOLS
will depend on !MODPROBE_SMALL.

The try_to_mmap_module() function is not called in modprobe-small.c,
so I will let FEATURE_INSMOD_TRY_MMAP depend on !MODPROBE_SMALL for
now.

Signed-off-by: Kang-Che Sung <explorer09@gmail.com>
Likewise, FEATURE_2_4_MODULES is not used by modprobe-small.

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

wget: add a big explanation what TLS code implements and what does not

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

libbb: match_fstype() is unreadable in the extreme, fixing it

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

vi,fsck: do not use build timestamp unconditionally. Closes 9626

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

taskset: separate "current" and "new" strings

Better for constant sharing:

   text    data     bss     dec     hex filename
 912997     485    6856  920338   e0b12 busybox_old
 912988     485    6856  920329   e0b09 busybox_unstripped

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

typo in comment

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

taskset: rewrite to be task size-agnostic

function                                             old     new   delta
packed_usage                                       31130   31190     +60
taskset_main                                         623     525     -98
------------------------------------------------------------------------------
(add/remove: 0/0 grow/shrink: 1/1 up/down: 60/-98)            Total: -38 bytes

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

*: add comment about APPLET_ODDNAME format

It confused me more than once

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

httpd: defend against attempts to OOM us. Closes 9611

We were strdup'ing "Cookie: foo" every time we saw it.

function                                             old     new   delta
handle_incoming_and_exit                            2733    2821     +88

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

adduser: help for -G GRP is misleading, fix it

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

sh: fix FEATURE_SH_STANDALONE help text: it does not enable NOFORK

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

libbb: spawn_and_wait() fflushes before forking NOEXEC; child reinits logmode

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

link: new applet

coreutils grew itself a tiny simplistic alternative to ln:

Usage: link FILE LINK

Create hard LINK to FILE

function                                             old     new   delta
link_main                                              -      75     +75
packed_usage                                       31114   31131     +17
applet_names                                        2564    2569      +5
applet_main                                         1480    1484      +4
applet_install_loc                                   185     186      +1
------------------------------------------------------------------------------
(add/remove: 1/0 grow/shrink: 4/0 up/down: 102/0)             Total: 102 bytes

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

xxd: make -p output lines actually end with a newline

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

xxd: implement -p

While at it, tweaked hexdump --help

function                                             old     new   delta
xxd_main                                             364     414     +50
packed_usage                                       31097   31114     +17

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

libbb: shrink sump.c

function                                             old     new   delta
dot_flags_width_chars                                  -      16     +16
int_convs                                              -       7      +7
lcc                                                    7       -      -7
index_str                                             16       -     -16
rewrite                                             1013     937     -76
------------------------------------------------------------------------------
(add/remove: 2/2 grow/shrink: 0/1 up/down: 23/-99)            Total: -76 bytes

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

ftpd/ls: show directories first

Old TODO finally done

function                                             old     new   delta
ls_main                                              548     568     +20
packed_usage                                       31116   31097     -19

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

hexdump/xxd: a bit more condensed formats

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

xxd: new applet

Yet Another Hexdumper

function                                             old     new   delta
xxd_main                                               -     364    +364
packed_usage                                       31046   31116     +70
applet_names                                        2560    2564      +4
applet_main                                         1476    1480      +4
rewrite                                             1022    1013      -9
------------------------------------------------------------------------------
(add/remove: 2/0 grow/shrink: 3/1 up/down: 442/-9)            Total: 433 bytes

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

ftpd: new option -a ANON_USER to allow anonymous logins

Anonymous ftpd login is useful even when ftpd authentication feature
is enabled. Anonymous logins provide simple password-less connection
for FTP clients.

To allow password-less connection user command line option '-a USER' is
added. This option specifies the system user to use when
'anonymous' username is given in USER command. No password is required
in this case.

function                                             old     new   delta
ftpd_main                                           2164    2232     +68
packed_usage                                       31015   31046     +31
------------------------------------------------------------------------------
(add/remove: 0/0 grow/shrink: 2/0 up/down: 99/0)               Total: 99 bytes

Signed-off-by: Andrey Mozzhuhin <amozzhuhin@yandex.ru>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

wget: add support for -S --server-response

Based on the patch by stephane.billiart@gmail.com

function                                             old     new   delta
ftpcmd                                                87     129     +42
fgets_and_trim                                        86     119     +33
static.wget_longopts                                 234     252     +18
packed_usage                                       31002   31015     +13
wget_main                                           2535    2540      +5
gethdr                                               158     163      +5
retrieve_file_data                                   424     428      +4
------------------------------------------------------------------------------
(add/remove: 0/0 grow/shrink: 7/0 up/down: 120/0)             Total: 120 bytes

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

remove "local" bashism from a few scripts

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

tls: can download kernels now :)

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

tls: if got CERTIFICATE_REQUEST, send an empty CERTIFICATE

wolfssl test server is not satisfied by an empty one,
but some real servers might be.

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

tls: add 2nd cipher_id, TLS_RSA_WITH_AES_128_CBC_SHA, so far it doesn't work

Good news that TLS_RSA_WITH_AES_256_CBC_SHA256 still works with new code ;)

This change adds inevitable extension to have different sized hashes and AES key sizes.
In libbb, md5_end() and shaX_end() are extended to return result size instead of void -
this helps *a lot* in tls (the cost is ~5 bytes per _end() function).

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

ls: convert DISP_DIRNAME to a bool variable

function                                             old     new   delta
ls_main                                              553     548      -5

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

ls: get rid of opt_flags[], handle -l1c through option_mask32

function                                             old     new   delta
display_single                                       885     888      +3
scan_and_display_dirs_recur                          496     486     -10
opt_flags                                             11       -     -11
ls_main                                              618     553     -65
------------------------------------------------------------------------------
(add/remove: 0/1 grow/shrink: 1/2 up/down: 3/-86)             Total: -83 bytes

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

ls: handle -x through option_mask32, remove default -C from --help

function                                             old     new   delta
packed_usage                                       31024   31002     -22

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

ls: handle -d and -R through option_mask32

function                                             old     new   delta
scan_and_display_dirs_recur                          545     550      +5
display_single                                      1039    1044      +5
append_char                                           67      68      +1
display_files                                        399     396      -3
ls_main                                              736     717     -19
opt_flags                                             68      11     -57
.rodata                                           168864  168784     -80
------------------------------------------------------------------------------
(add/remove: 0/0 grow/shrink: 3/4 up/down: 11/-159)          Total: -148 bytes

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

ls: handle -p and -F through option_mask32

function                                             old     new   delta
display_single                                      1039    1044      +5
append_char                                           67      68      +1
display_files                                        399     396      -3
scan_and_display_dirs_recur                          545     541      -4
ls_main                                              736     722     -14
------------------------------------------------------------------------------
(add/remove: 0/0 grow/shrink: 2/3 up/down: 6/-21)             Total: -15 bytes

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

ls: LIST_ID_NAME/ID_NUMERIC/LOPT/LONG are the same, merge as LONG

Also, handle -n and --full-time using option_mask32

function                                             old     new   delta
scan_and_display_dirs_recur                          548     545      -3
display_single                                      1053    1039     -14
ls_main                                              777     736     -41
------------------------------------------------------------------------------
(add/remove: 0/0 grow/shrink: 0/3 up/down: 0/-58)             Total: -58 bytes

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

ls: make -Z output compatible with GNU coreutils 8.25

This also makes it clear that LIST_MODEBITS == LIST_LOPT

function                                             old     new   delta
ls_main                                              752     777     +25
display_files                                        400     399      -1
display_single                                      1055    1053      -2
my_stat                                              320     317      -3
packed_usage                                       32311   32280     -31
.rodata                                           168895  168864     -31
------------------------------------------------------------------------------
(add/remove: 0/0 grow/shrink: 1/5 up/down: 25/-68)            Total: -43 bytes

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

ls: LIST_NLINKS/SIZE/DATE_TIME/SYMLINK are always the same, merge as LIST_LOPT

They are always set and cleared together

function                                             old     new   delta
append_char                                           69      68      -1
display_single                                       910     908      -2
ls_main                                              677     664     -13
------------------------------------------------------------------------------
(add/remove: 0/0 grow/shrink: 0/3 up/down: 0/-16)             Total: -16 bytes

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

ls: handle -s through option_mask32

function                                             old     new   delta
scan_and_display_dirs_recur                          493     494      +1
sort_and_display_files                               424     423      -1
ls_main                                              682     677      -5

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

ls: handle -i through option_mask32

function                                             old     new   delta
sort_and_display_files                               420     424      +4

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

ls: handle -a and -A through option_mask32

function                                             old     new   delta
ls_main                                              685     682      -3

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

ls: handle all sort options through option_mask32

function                                             old     new   delta
packed_usage                                       31007   31024     +17
my_stat                                              318     324      +6
sort_and_display_files                               419     420      +1
print_name                                           216     217      +1
sortcmp                                              228     216     -12
opt_flags                                             96      64     -32
ls_main                                              732     685     -47
------------------------------------------------------------------------------
(add/remove: 0/0 grow/shrink: 4/3 up/down: 25/-91)            Total: -66 bytes

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

tls: reorder tls_handshake_data fields for smaller size, tweak comments

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

tls: send EMPTY_RENEGOTIATION_INFO_SCSV in our client hello

Hoped this can make cdn.kernel.org to like us more. Nope.
While at it, made error reporting more useful.

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

tls: set TLS_DEBUG to 0; placate a gcc indentation warning

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

separate TLS code into a library, use in in wget

A new applet, ssl_client, is the TLS debug thing now.
It doubles as wget's NOMMU helper.
In MMU mode, wget still forks, but then directly calls TLS code,
without execing.

This can also be applied to sendmail/popmail (SMTPS / SMTP+starttls support)
and nc --ssl (ncat, nmap's nc clone, has such option).

function                                             old     new   delta
tls_handshake                                          -    1691   +1691
tls_run_copy_loop                                      -     443    +443
ssl_client_main                                        -     128    +128
packed_usage                                       30978   31007     +29
wget_main                                           2508    2535     +27
applet_names                                        2553    2560      +7
...
xwrite_encrypted                                     360     342     -18
tls_main                                            2127       -   -2127
------------------------------------------------------------------------------
(add/remove: 4/1 grow/shrink: 13/8 up/down: 2351/-2195)       Total: 156 bytes

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

ls: -1 should be ignored by -l (and options which imply -l)

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

ls: more correct handling of -c, -u

function                                             old     new   delta
my_stat                                              302     318     +16
packed_usage                                       30977   30969      -8
display_single                                       928     910     -18
sortcmp                                              258     228     -30
ls_main                                              776     732     -44
------------------------------------------------------------------------------
(add/remove: 0/0 grow/shrink: 1/4 up/down: 16/-100)           Total: -84 bytes

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

ls: replace -e with --full-time, add --group-directories-first, delete -K

-K and -e were non-standard

function                                             old     new   delta
static.ls_longopts                                     9      47     +38
ls_main                                              748     776     +28
display_single                                       901     928     +27
sortcmp                                              254     258      +4
ls_options                                            32      31      -1
opt_flags                                            100      96      -4
packed_usage                                       31032   30977     -55
------------------------------------------------------------------------------
(add/remove: 0/0 grow/shrink: 4/3 up/down: 97/-60)             Total: 37 bytes

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

ip: better --help

Was:
    Usage: ip [OPTIONS] address|route|link|tunnel|neigh|rule [COMMAND]

    ip [OPTIONS] OBJECT [COMMAND]
    where OBJECT := address|route|link|tunnel|neigh|rule
    OPTIONS := -f[amily] inet|inet6|link | -o[neline]

User: instead of repeating list of OBJECTs twice, you could at least
show available COMMANDs...

Now:
    Usage: ip [OPTIONS] address|route|link|tunnel|neigh|rule [COMMAND]

    OPTIONS := -f[amily] inet|inet6|link | -o[neline]
    COMMAND :=
    ip addr add|del IFADDR dev IFACE | show|flush [dev IFACE] [to PREFIX]
    ip route list|flush|add|del|change|append|replace|test ROUTE
    ip link set IFACE [up|down] [arp on|off] | show [IFACE]
    ip tunnel add|change|del|show [NAME]
        [mode ipip|gre|sit]
        [remote ADDR] [local ADDR] [ttl TTL]
    ip neigh show|flush [to PREFIX] [dev DEV] [nud STATE]
    ip rule [list] | add|del SELECTOR ACTION

While at it, tweak tc --help too (it stays disabled, thus no effect)

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

more ip --help fixes

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

make --help texts smaller

function                                             old     new   delta
packed_usage                                       31035   30968     -67

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

make --help texts more uniform

function                                             old     new   delta
packed_usage                                       31062   31035     -27

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

tls: reorder tls_state fields for smaller offsets

function                                             old     new   delta
xwrite_encrypted                                     363     360      -3
xwrite_and_update_handshake_hash                     117     114      -3
tls_xread_handshake_block                             72      69      -3
tls_error_die                                        211     202      -9
tls_get_outbuf                                        64      49     -15
tls_main                                            2163    2127     -36
tls_xread_record                                     702     639     -63
------------------------------------------------------------------------------
(add/remove: 0/0 grow/shrink: 0/7 up/down: 0/-132)           Total: -132 bytes

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

tls: send SNI in the client hello

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

tls: check size on "MAC-only, no crypt" code path too

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

tls: AES decrypt does one unnecessary memmove

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

tls: make input buffer grow as needed

As it turns out, it goes only up to "inbuf_size:4608"
for kernel.org - fixed 18kb buffer was x4 larger than necessary.

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

password utils: improve --help, make DEFAULT_PASSWD_ALGO visible if CHPASSWD

Was:
    $ cryptpw --help
    ...
    Print crypt(3) hashed PASSWORD

        -P,--password-fd=N Read password from fd N
        -m,--method=TYPE Encryption method
        -S,--salt=SALT

User: "What methods exist? which one os default?"

Now:
    Print crypt(3) hashed PASSWORD

        -P,--password-fd N Read password from fd N
        -m,--method TYPE des,md5,sha256/512 (default des)
        -S,--salt SALT

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

tls: improve i/o loop

With tls_has_buffered_record(), entire kernel.org response
is printed at once, without 6 second pause to see its delayed EOF.

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

tls: was psAesDecrypt'ing one block too many, trashing buffered data

For the first time

printf "GET / HTTP/1.1\r\nHost: kernel.org\r\n\r\n" | ./busybox tls kernel.org

successfully reads entire server response and TLS shutdown.

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

rdate: time(NULL) is shorter than time(&var)

function                                             old     new   delta
rdate_main                                           251     246      -5

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

rdate: make it do something remotely sane, facing 32-bit time overflow

function                                             old     new   delta
rdate_main                                           251     254      +3
packed_usage                                       31029   31023      -6

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

tls: do not use common_bufsiz

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

tls: decode alerts and in particular, EOF alert.

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

tls: add the i/o loop - largish rework of i/o buffering

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

tls: fix ROL/ROR x86 optimization

ALWAYS_INLINE:

function                                             old     new   delta
psAesInitKey                                         825     824      -1
ROR                                                    5       -      -5
setup_mix2                                           148     134     -14
psAesDecryptBlock                                   1184    1139     -45
psAesEncryptBlock                                   1193    1102     -91
------------------------------------------------------------------------------
(add/remove: 0/1 grow/shrink: 0/4 up/down: 0/-156)           Total: -156 bytes

ALWAYS_INLINE + __builtin_constant_p(shift_cnt):

function                                             old     new   delta
ROR                                                    5       -      -5
psAesInitKey                                         825     818      -7
setup_mix2                                           148     123     -25
psAesDecryptBlock                                   1184    1078    -106
psAesEncryptBlock                                   1193    1017    -176
------------------------------------------------------------------------------
(add/remove: 0/1 grow/shrink: 0/4 up/down: 0/-319)           Total: -319 bytes

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

tls: tested PSTM_X86_64, not enabling it - too large

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

tls: commented out psPool_t use

function                                             old     new   delta
psAesEncrypt                                         159     162      +3
der_binary_to_pstm                                    42      40      -2
xwrite_and_hash                                      437     434      -3
xread_tls_block                                      446     443      -3
pstm_div_2d                                          449     444      -5
psAesDecrypt                                         179     174      -5
pstm_init_size                                        52      45      -7
pstm_init                                             46      39      -7
pstm_to_unsigned_bin                                 165     157      -8
tls_main                                            1265    1256      -9
pstm_mulmod                                          132     123      -9
pstm_mod                                             125     116      -9
pstm_init_copy                                        93      84      -9
psAesInitKey                                         840     825     -15
send_client_key_exchange                             362     342     -20
psAesInit                                            103      80     -23
psRsaEncryptPub                                      429     403     -26
psAesDecryptBlock                                   1211    1184     -27
psAesEncryptBlock                                   1223    1193     -30
pstm_exptmod                                        1582    1524     -58
pstm_div                                            1557    1472     -85
------------------------------------------------------------------------------
(add/remove: 0/0 grow/shrink: 1/20 up/down: 3/-360)          Total: -357 bytes

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

ntpd: print result of hostname resolution

This is particularly useful if hostname resolution is triggered by
host non-reachability: I saw this in real-life, without the message
it is not at all obvious that IP that we use for a specific host
has changed.

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

fdisk: add typical values of -H and -S to --help

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

tls: teach it to decrypt AES256-encrypted data

This adds decryption only.
There is no MAC verification, code simply throws away MAC.

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

tls: trim comments

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

tls: teach it to send AES256-encrypted data

>> CLIENT_HELLO
wrote 50 bytes
insize:0 tail:0
got block len:74
got HANDSHAKE
<< SERVER_HELLO
insize:79 tail:0
got block len:2397
got HANDSHAKE
<< CERTIFICATE
key bytes:271, first:0x00
server_rsa_pub_key.size:256
insize:2402 tail:0
got block len:4
got HANDSHAKE
<< SERVER_HELLO_DONE
>> CLIENT_KEY_EXCHANGE
wrote 267 bytes
master secret:c51df5b1e3b3f57373cdd8ea28e8ce562059636cf9f585d0b89c7f4bacec97e674d7b91f93e7b500cb64637f240c3b78
client_write_MAC_key:3b0b7e2bab241b629c37eb3a3824f09b39fe71a00876b0c8026dda16ef0d2f82
client_write_key:d36e801470ed2f0a8fc886ac25df57ffbe4265d06e3192122c4ef4df1e32fab2
>> CHANGE_CIPHER_SPEC
from secret: c51df5b1e3b3f57373cdd8ea28e8ce562059636cf9f585d0b89c7f4bacec97e674d7b91f93e7b500cb64637f240c3b78
from labelSeed: 636c69656e742066696e6973686564b22e0e6008b8ee218cc02e4a93e4a42b570535f9b57662e262d43b379d125b69
=> digest: a45bfee8ed6507a2a9920d0c
>> FINISHED
before crypt: 5 hdr + 16 data + 32 hash bytes
writing 5 + 16 IV + 64 encrypted bytes, padding_length:0x0f
wrote 85 bytes
insize:9 tail:0
got block len:1
<< CHANGE_CIPHER_SPEC
insize:6 tail:0
got block len:80
< hdr_type:22 ver:3.3 len:80 type:21 len24:9541723 |1591985b...a3da|

The last line is the server's FINISHED response, encrypted.

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

tls: added AES code and made it compile. not used yet

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

tls: massage writing for encryption support; finer-grained debug

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

tls: address one easy FIXME, tidy up comments

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

tls: process CHANGE_CIPHER_SPEC and FINISHED from server

Successfully finishes handshake with test servers using NULL-SHA256
cipher.

The "only" thing remaining before there is a chance
this can actually work with real servers is AES encrypt/decrypt.

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

tls: rearrange function order, improve comments

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

tls: make our send_client_finished() pass server check

sha256 hash should be calculated over incoming handshake packets too!

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

tls: format FINISHED message properly for unencrypted, but sha256 signed mode

Now it at least looks correct, but unfortunately "openssl s_server"
says my hash is wrong.

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

Assorted warning fixes and added a comment, no code changes

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>