The unsigned char buffer[] must have size of (at least)
SIZE_GET_SLOT_STATUS.

Thanks to Ran Benita for the patch
" Avoid LIBUSB_ERROR_OVERFLOW errors #164 "
https://github.com/LudovicRousseau/CCID/pull/164

Signed-off-by: Ludovic Rousseau <ludovic.rousseau@free.fr>

The Protocol Data Structure is:
- 5 bytes long for a T=0 card
- 7 bytes long for a T=1 card

It may happen that the driver sends parameters for a T=0 card (5 bytes)
but the reader answers with a T=1 buffer (7 bytes).
The buffer is then too small for libusb and we get a LIBUSB_ERROR_OVERFLOW
error.

This should not happen in normal use but may happen with my HandlerTest
tool when I insert a T=1 card and ask for a T=0 communication. Some
reader (like the Cherry Smart Terminal 1150) may answer with a T=1
Protocol Data Structure.

This happened on composite devices with 2 or more CCID interfaces.

When the device is removed the first interface is correctly stopped but
the driver and pcsc-lite are still trying to use the other CCID
interfaces. We then get libusb errors:
00000025 [139657349199552] ../src/ccid_usb.c:1614:InterruptRead() libusb_submit_transfer failed: LIBUSB_ERROR_NO_DEVICE

When a device is removed (unplugged) we must mark disconnected all the
CCID interfaces and not just all the slots.

On a composite device (2 or more CCID interfaces) the driver was still
trying to talk to the second (or more) CCID interface and generated
libusb errors LIBUSB_ERROR_NO_DEVICE.

The only change is the firmware release from 1.01 to 1.02.

Fix build error:
<command-line>: error: "_FORTIFY_SOURCE" redefined [-Werror]

CCID defines another type of message type on the interrupt pipe,
HardwareError, and others may also be conceivably sent. So check the
message type instead of assuming it's a NotifySlotChange.

While at it, I added debug logging of HardwareError, it might be useful
for someone.

This is somewhat nice for readability, but will be useful for the next
commit.

The `HAVE_` defines for these functions are not used.

This brings parity with the meson build.

It is present in configure.ac but not in meson, so meson builds end up
using the fallback.

Ref 09c8471f.

Use "meson setup builddir -Denable-extras=true" to also build:
- examples/scardcontrol
- contrib/Kobil_mIDentity_switch
- contrib/RSA_SecurID

Thanks to Martin Paljak for the idea
" Do not build contrib and examples by default but require #4 "
https://github.com/LudovicRousseau/CCID/pull/4

Maksim Ivanov authored 5 months ago and Ludovic Rousseau committed 5 months ago

It seems generally useful to mark "local" things as static, so that they
can't cause symbol collision between different source files.

We now have:
$ nm -g --defined-only libccid.so
000000000000bf90 T IFDHCloseChannel
000000000000ea40 T IFDHControl
000000000000bf80 T IFDHCreateChannel
000000000000bf70 T IFDHCreateChannelByName
0000000000010ad0 T IFDHGetCapabilities
0000000000010880 T IFDHICCPresence
000000000000c1b0 T IFDHPowerICC
000000000000c030 T IFDHSetCapabilities
000000000000c5d0 T IFDHSetProtocolParameters
000000000000e260 T IFDHTransmitToICC

Thanks to Maksim Ivanov for the idea
" Mark suitable variables and functions static #147 "
https://github.com/LudovicRousseau/CCID/pull/147

../src/commands.c: In function ‘CmdEscapeCheck’:
../src/commands.c:1057:48: warning: ‘old_read_timeout’ may be used uninitialized [-Wmaybe-uninitialized]
 1057 |                 ccid_descriptor -> readTimeout = old_read_timeout;
      |                                                ^
../src/commands.c:948:13: note: ‘old_read_timeout’ was declared here
  948 |         int old_read_timeout;
      |             ^

Thanks to Andrew Baxter for the bug report
" LIBUSB_ERROR_TIMEOUT on first connect #206 "
fixes: https://github.com/LudovicRousseau/PCSC/issues/206

Signed-off-by: Ludovic Rousseau <ludovic.rousseau@free.fr>

Fixes:
NOTICE: Future-deprecated features used:
 * 1.1.0: {'"boolean option" keyword argument "value" of type str'}

Thanks to https://github.com/mesonbuild/meson/discussions/11254#discussioncomment-4622562

Fix link issue on Ubuntu 20.04

$ ninja
[32/33] Linking target parse.
FAILED: parse
cc  -o parse 'parse@exe/meson-generated_tokenparser.c.o' 'parse@exe/src_parse.c.o' 'parse@exe/src_debug.c.o' 'parse@exe/src_ccid_usb.c.o' 'parse@exe/src_sys_unix.c.o' 'parse@exe/src_strlcpy.c.o' 'parse@exe/src_simclist.c.o' -Wl,--as-needed -Wl,--no-undefined -Wl,--start-group /usr/lib/x86_64-linux-gnu/libusb-1.0.so /usr/lib/x86_64-linux-gnu/libz.so -Wl,--end-group
/usr/bin/ld: parse@exe/src_ccid_usb.c.o: undefined reference to symbol 'pthread_join@@GLIBC_2.2.5'
/usr/bin/ld: /lib/x86_64-linux-gnu/libpthread.so.0: error adding symbols: DSO missing from command line
collect2: error: ld returned 1 exit status
[33/33] Generating Info.plist with a meson_exe.py custom command.
ninja: build stopped: subcommand failed.

Thanks to Peter Helcmanovsky for the bug report and patch.

This explicitly reverts commit eec7cdf0
because it was a bad idea.

The motivating bug report was https://github.com/LudovicRousseau/PCSC/issues/124
and the issue there occurred when building from a git clone, running
./bootstrap && ./configure && make, and having:

- configure succeed
- make "succeeeds" at having $LEX run, do nothing and fail to generate
  required sources
- compiling nonexistent files fail with highly confusing errors

The autoconf manual has always documented the correct way to handle this
is to check if lex is unavailable, and set it to the famous automake
wrapper "missing", which checks if a program is missing at build time
rather than at ./configure time, and fails the build if the rule cannot
be run. This means:

When building from a git clone, if flex is not available then
- configure succeeds
- make fails to run $LEX, and tells you to install flex

The previous attempt to fix the highly confusing error instead resulted
in configure erroring out, and saying flex is required, even when it is
*not* required because a `make dist` tarball was used, which contains
pregenerated tokenparser.c for the express purpose of making flex
unnecessary.

See autoconf documentation on $LEX:
https://www.gnu.org/software/autoconf/manual/autoconf-2.72/html_node/Particular-Programs.html#index-AC_005fPROG_005fLEX-1

And automake documentation on why to use "missing":
https://www.gnu.org/software/automake/manual/html_node/maintainer_002dmode.html

The error was:
[30/32] Linking target parse
FAILED: parse
cc  -o parse parse.p/meson-generated_tokenparser.c.o parse.p/src_parse.c.o parse.p/src_debug.c.o parse.p/src_ccid_usb.c.o parse.p/src_sys_unix.c.o parse.p/src_simclist.c.o -Wl,--as-needed -Wl,--no-undefined -Wl,--start-group /usr/lib/x86_64-linux-gnu/libusb-1.0.so /usr/lib/x86_64-linux-gnu/libz.so -Wl,--end-group
/usr/bin/ld: parse.p/src_debug.c.o: in function `log_xxd':
/home/user/workspace/CCID/builddir/../src/debug.c:221: undefined reference to `strlcpy'
collect2: error: ld returned 1 exit status

Thanks to Jahns Ralf for the bug report.

./GetSetNAD.py:35: SyntaxWarning: "is" with a literal. Did you mean "=="?
  if status is 0:
./GetSetNAD.py:51: SyntaxWarning: "is" with a literal. Did you mean "=="?
  if status is 0: