Added Linux support to qc_handle_{accept,bind,connect} functions. #23

MCApollo · 2020-08-06T20:16:42Z

Handles issue #22, open to changes.

EDIT

Help Wanted:
tunnel out:ip:port returns a deadlock error:
accept failed: 35 - I can't get a breakpoint to hit.

sickcodes · 2020-08-07T01:22:03Z

Builds on both Linux and OSX
On the first try, it worked, it locked up the machine but the port was definitely open.

Then I built the new tcp tunnel on OSX with the new qemu and I couldn't SSH in on OSX.

Brought the hfs back into linux and SSH tunnel fails, but fails to connect. It doesn't give the 255 error like before.

I'll try again shortly

but stalls without adding the new tcp-tunnel from xnu-qemu-arm64-tools.

export XNU_SOURCES=$PWD/darwin-xnu
export KERNEL_SYMBOLS_FILE=$PWD/symbols.nm
export QEMU_DIR=$PWD/xnu-qemu-arm64/
export NUM_BLOCK_DEVS=2
export KERNEL_CACHE=$PWD/kernelcache.release.n66.out
export DTB_FIRMWARE=$PWD/Firmware/all_flash/DeviceTree.n66ap.im4p.out
export DRIVER_FILENAME=$PWD/aleph_bdev_drv.bin
export IOS_DIR=$PWD
export QEMU_DIR=$PWD/xnu-qemu-arm64/
export HFS_MAIN=$PWD/hfs.main
export HFS_SEC=$PWD/hfs.sec

cd xnu-qemu-arm64

# guest cross-compilers enabled: aarch64-linux-gnu-gcc
./configure \
--target-list=aarch64-softmmu \
--disable-capstone \
--disable-werror \
--disable-pie \

xnu-qemu-arm64/aarch64-softmmu/qemu-system-aarch64 \
-M iPhone6splus-n66-s8000,kernel-filename=${KERNEL_CACHE},dtb-filename=${DTB_FIRMWARE},driver-filename=${DRIVER_FILENAME},qc-file-0-filename=${HFS_MAIN},qc-file-1-filename=${HFS_SEC},kern-cmd-args="debug=0x8 kextlog=0xfff cpus=1 rd=disk0 serial=2",xnu-ramfb=off \
-cpu max \
-m 6G \
-serial mon:stdio \
-vga std \




export PATH=$PATH:/iosbinpack64/usr/bin:/iosbinpack64/bin:/iosbinpack64/usr/sbin:/iosbinpack64/sbin:/sbin:/bin:/usr/sbin:/usr/bin

MCApollo · 2020-08-07T01:49:16Z

In the tcp-tunnel’s Makefile, uncomment the part about FLAG += -DLINUX; Make sure the filesystem is non-dirty before transferring or use hfsprogs to check the filesystem.

No real easy way to have a ‘universal’ binary, EDEADLK is 11 on sys/errno.h on Darwin- replacing EAGAIN.

If you’ve verified issues on macos with lldb, let me know. Nothing was changed for macos except making structs for sockaddr a alias they were originally were.

sickcodes · 2020-08-07T01:55:36Z

Alright, will try now.

If I use the non-patched tcp-tunnel, I can almost fully connect:

$ ssh -vvvvv localhost -p2222
OpenSSH_8.3p1, OpenSSL 1.1.1g  21 Apr 2020
debug1: Reading configuration data /home/user/.ssh/config
debug1: /home/user/.ssh/config line 1: Applying options for *
debug1: Reading configuration data /etc/ssh/ssh_config
debug2: resolving "localhost" port 2222
debug2: ssh_connect_direct
debug1: Connecting to localhost [::1] port 2222.
debug1: connect to address ::1 port 2222: Connection refused
debug1: Connecting to localhost [127.0.0.1] port 2222.
debug1: Connection established.
debug1: identity file /home/user/.ssh/id_rsa type 0
debug1: identity file /home/user/.ssh/id_rsa-cert type -1
debug1: identity file /home/user/.ssh/id_dsa type -1
debug1: identity file /home/user/.ssh/id_dsa-cert type -1
debug1: identity file /home/user/.ssh/id_ecdsa type -1
debug1: identity file /home/user/.ssh/id_ecdsa-cert type -1
debug1: identity file /home/user/.ssh/id_ecdsa_sk type -1
debug1: identity file /home/user/.ssh/id_ecdsa_sk-cert type -1
debug1: identity file /home/user/.ssh/id_ed25519 type -1
debug1: identity file /home/user/.ssh/id_ed25519-cert type -1
debug1: identity file /home/user/.ssh/id_ed25519_sk type -1
debug1: identity file /home/user/.ssh/id_ed25519_sk-cert type -1
debug1: identity file /home/user/.ssh/id_xmss type -1
debug1: identity file /home/user/.ssh/id_xmss-cert type -1
debug1: Local version string SSH-2.0-OpenSSH_8.3
kex_exchange_identification: Connection closed by remote host
Connection closed by 127.0.0.1 port 2222

# unpatched version btw

Trying your tips now

My HFS is always dirty on Linux, I'll check it out, does it get dirty if I boot it on OSX first?

sickcodes · 2020-08-07T02:00:01Z

FYI, trying your steps now

### patched tcp-tunnel
$ fsck.hfsplus hfs_new.main
** hfs_new.main
   Executing fsck_hfs (version 540.1-Linux).
** Checking non-journaled HFS Plus Volume.
** Detected a case-sensitive volume.
   The volume name is PeaceB16B92.arm64UpdateRamDisk
** Checking extents overflow file.
** Checking catalog file.
** Checking multi-linked files.
** Checking catalog hierarchy.
** Checking extended attributes file.
** Checking volume bitmap.
** Checking volume information.
** The volume PeaceB16B92.arm64UpdateRamDisk appears to be OK.

### upatched tcp-tunnel
$ fsck.hfsplus hfs.main
** hfs.main
   Executing fsck_hfs (version 540.1-Linux).
** Checking non-journaled HFS Plus Volume.
** Detected a case-sensitive volume.
   The volume name is PeaceB16B92.arm64UpdateRamDisk
** Checking extents overflow file.
** Checking catalog file.
** Checking multi-linked files.
** Checking catalog hierarchy.
** Checking extended attributes file.
** Checking volume bitmap.
** Checking volume information.
** The volume PeaceB16B92.arm64UpdateRamDisk appears to be OK.

sickcodes · 2020-08-07T02:33:56Z

Okay I think it got dirty when I rsynced it from the KVM guest to host (compressed), about to test!

sickcodes · 2020-08-07T03:02:36Z

@MCApollo amazing dude!

-bash-4.4# uname -a
Darwin localhost 18.2.0 Darwin Kernel Version 18.2.0: Tue Oct 16 21:02:23 PDT 2018; root:xnu-4903.222.5~1/RELEASE_ARM64_S8000 iPhone8,2

Will test the other way to make sure

sickcodes · 2020-08-08T02:18:54Z

Handles issue #22, open to changes.

EDIT

Help Wanted:

tunnel out:ip:port returns a deadlock error:

accept failed: 35 - I can't get a breakpoint to hit.

I get the same error, unless I tunnel in. I tried to tunnel in

sshpass -p "alpine" ssh -N [email protected] -p 2222 5900:127.0.0.1:5900

Was trying to run VNC, signed screendumpd launch daemon with jtool and I can reach the VNC server and get a response via loop back tunneling except something about the VNC doesn't work.

I cant tunnel out I also get error 35, I can only have one connection at a time.

The VNC drops and the error is BigEndian = 0 or something. Can't you tunnel everything out on the one port, like a VPN?

Whenever a connection drops, I can't use that host port again

I opened
tunnel 59001:0.0.0.0:5900

Everything works and I can get thru ssh tunneled in, and then i get error 61, which I assume is Refused Connection in the console

aronsky · 2020-08-09T10:43:44Z

Hi @MCApollo

First of all, thanks for the PR! It's great to see new features such as Linux support being added to the project! However, the extensive use of ifdefs and define wrappers around various structs really hurts readability, IMO. I suggest refactoring this by wrapping the cpu_memory_rw_debug to access the struct fields correctly. This will minimize the number of changes required, and make them easier to review. We basically need something similar to ntohs/htons, but for sockaddr in place of short, and Linux/Darwin in place of host/network byte order. I'll try to come up with a suggested implementation of such a wrapper soon.

MCApollo · 2020-08-09T16:26:26Z

That’s a actually a better idea and was hoping someone would help with; Feel free to continue to use this PR, I know this code is god-awful and is a straight up bandaid until someone/I has the time.

I can continue to work on this, been busy with some other things.

There’s really two issues that need a checkbox

Darwin uses sin_len and may slightly deviant from the network standard. addrlen is the same by standard.
Darwin uses 35 as a EAGAIN, maybe it would be better to edit the code on the host side on accept to report EAGAIN as 35.

Some notes are:

SOCK_NONBLOCK is needed on Linux or any connection hangs on accept. I’m unsure if this is the default on Darwin or I’m missing something.
Not giving addrlen a value in accept causes it fail.

aronsky · 2020-08-11T06:34:51Z

hw/arm/guest-socket.c

+#ifdef __APPLE__
+# define ADDR addr
+# undef NETWORK_HACK
+#else /* Linux, etc. */
+# define ADDR darwin_addr
+# define NETWORK_HACK 1
+#endif


There's no reason for the NETWORK_HACK and ADDR defines, we can condition the functions based on __APPLE__ only.

aronsky · 2020-08-11T06:35:16Z

hw/arm/guest-socket.c

+#endif
+
+static int convert_error(int err) {
+#ifndef NETWORK_HACK


Suggested change

#ifndef NETWORK_HACK

#ifdef __APPLE__

aronsky · 2020-08-11T06:36:13Z

hw/arm/guest-socket.c

+# define NETWORK_HACK 1
+#endif
+
+static int convert_error(int err) {


I'd suggest a function name that makes the platform clear, such as darwin_error.

aronsky · 2020-08-11T06:41:35Z

hw/arm/guest-socket.c

+#endif
+}
+
+static struct sockaddr_in convert_sockaddr_to(S_SOCKADDR_IN from) {


Notice that you only call this function when NETWORK_HACK is defined, which means the following:

The #ifndef inside the function is redundant

The name of the function can be more explicit, such as convert_sockaddr_darwin_to_gnu

It should receive darwin_sockaddr_in explicitly (and the S_SOCKADDR_IN define can be dropped)

Suggested change

static struct sockaddr_in convert_sockaddr_to(S_SOCKADDR_IN from) {

static struct sockaddr_in convert_sockaddr_darwin_to_gnu(darwin_sockaddr_in from) {

aronsky · 2020-08-11T06:41:54Z

hw/arm/guest-socket.c

+#ifndef NETWORK_HACK
+    return from;
+#else


Redundant...

aronsky · 2020-08-11T07:27:14Z

hw/arm/guest-socket.c

+#ifdef NETWORK_HACK
+        S_SOCKADDR_IN ADDR = convert_sockaddr_from(addr);
+#endif
+        cpu_memory_rw_debug(cpu, (target_ulong) g_addr, (uint8_t*) &ADDR,
+                            sizeof(ADDR), 1);


Suggested change

#ifdef NETWORK_HACK

S_SOCKADDR_IN ADDR = convert_sockaddr_from(addr);

#endif

cpu_memory_rw_debug(cpu, (target_ulong) g_addr, (uint8_t*) &ADDR,

sizeof(ADDR), 1);

#ifndef __APPLE__

darwin_sockaddr_in darwin_addr = convert_sockaddr_gnu_to_darwin(addr);

cpu_memory_rw_debug(cpu, (target_ulong) g_addr, (uint8_t*) &darwin_addr,

sizeof(darwin_addr), 1);

#else

cpu_memory_rw_debug(cpu, (target_ulong) g_addr, (uint8_t*) &addr,

sizeof(addr), 1);

#endif

aronsky · 2020-08-11T07:28:21Z

hw/arm/guest-socket.c

+#ifdef NETWORK_HACK
+    S_SOCKADDR_IN ADDR;
+#endif


This will necessarily be darwin_sockaddr_in darwin_addr, if __APPLE__ is defined...

aronsky · 2020-08-11T07:30:21Z

hw/arm/guest-socket.c

+        cpu_memory_rw_debug(cpu, (target_ulong) g_addr, (uint8_t*) &ADDR,
+                            sizeof(ADDR), 0);

+#ifdef NETWORK_HACK
+        addr = convert_sockaddr_to(ADDR);
+#endif


Similar to the change in accept:

Suggested change

cpu_memory_rw_debug(cpu, (target_ulong) g_addr, (uint8_t*) &ADDR,

sizeof(ADDR), 0);

#ifdef NETWORK_HACK

addr = convert_sockaddr_to(ADDR);

#endif

#ifndef __APPLE__

cpu_memory_rw_debug(cpu, (target_ulong) g_addr, (uint8_t*) &darwin_addr,

sizeof(darwin_addr), 0);

addr = convert_sockaddr_to(darwin_addr);

#else

cpu_memory_rw_debug(cpu, (target_ulong) g_addr, (uint8_t*) &addr,

sizeof(addr), 0);

#endif

aronsky · 2020-08-11T07:31:59Z

hw/arm/guest-socket.c

+#ifdef NETWORK_HACK
+            ADDR = convert_sockaddr_from(addr);
+#endif
+            cpu_memory_rw_debug(cpu, (target_ulong) g_addr, (uint8_t*) &ADDR,
+                                sizeof(ADDR), 1);


Suggested change

#ifdef NETWORK_HACK

ADDR = convert_sockaddr_from(addr);

#endif

cpu_memory_rw_debug(cpu, (target_ulong) g_addr, (uint8_t*) &ADDR,

sizeof(ADDR), 1);

#ifndef __APPLE__

darwin_addr = convert_sockaddr_from(addr);

cpu_memory_rw_debug(cpu, (target_ulong) g_addr, (uint8_t*) &addr,

sizeof(addr), 1);

#else

cpu_memory_rw_debug(cpu, (target_ulong) g_addr, (uint8_t*) &addr,

sizeof(addr), 1);

#endif

aronsky · 2020-08-11T07:32:57Z

hw/arm/guest-socket.c

        }
    }

    return retval;
 }

-int32_t qc_handle_connect(CPUState *cpu, int32_t sckt, struct sockaddr *g_addr,
-                          socklen_t addrlen)
+int32_t qc_handle_connect(CPUState *cpu, int32_t sckt, S_SOCKADDR *g_addr,


Similar to bind...

MCApollo · 2020-08-11T08:37:17Z

Sounds about right, thanks for being better than me at C- I’ll work on finishing so we can finally fix this.

(TIL about compound literals for easier reading, 👍🏼)

As far as accept goes, I’m not sure if Darwin acts differently vs Linux (see github.com/dotnet/runtime/issues/25069#ref-commit-e36b3b7 -> Darwin inheriting flags from fd).

Correct me if I’m wrong, but my understanding is that this should set nonblock on the host side right? here:
out->fcntl(out->fd, F_SETFL, flags | O_NONBLOCK) < 0)
Without using accept4 to set nonblock, qemu obviously hangs waiting for a connection so your insight would be helpful.

aronsky · 2020-08-11T13:44:55Z

Darwin should be inheriting flags from the parent socket upon accept, similar to BSD, and contrary to Linux. We can explicitly set the flags on the newly accepted socket - this will be redundant on macOS, but it's not an expensive operation and will make sure the new socket is non-blocking as well, without conditionally calling accept on macOS and accept4 on Linux.

accept itself shouldn't be the blocking call, since we set the O_NONBLOCK flag beforehand here. The file descriptor it returns, however, is stored inside the in socket - while the only socket we set to be non-blocking inside handle_incoming_connection is the out socket (that's the line you linked in your comment). So what happens is that the accepted socket is non-blocking on macOS (because it inherits the flags we set on the listening socket), and everything works great, but on Linux, the socket is blocking, and blocks on recv...

I'd like to avoid conditional code as much as possible, which would leave the accept call here the same across Linux and macOS, and require the tunnel binary to explicitly set the non-blocking flag on the socket returned by accept. Alternatively, if we want to properly emulate the behavior of accept, and abstract away the inconsistencies between Linux and macOS (after all, the tcp-tunnel binary shouldn't really care what host it runs on), we can add an #ifdefed block that would copy the flags from the parent socket to the socket returned by accept (note, that if we go the second route - we should copy the flags instead of using accept4 with a hardcoded O_NONBLOCK, since we're emulating the behavior of Darwin, which simply inherits all the flags from the parent socket).

MCApollo · 2020-08-11T17:45:03Z

Yeah, the accept4 was a bandaid fix until I can confirm if it was a bug/“feature” since no documentation mentions it without digging through man pages.

I think forcing non_blocking breaks out anyway, so that’s next-up.

aronsky

Overall, looks good. We're close to the final revision. I still wish we could streamline the part where we copy the memory to/from the guest, but with pointers involved, I don't see a neat solution there.

hw/arm/guest-socket.c

aronsky · 2020-08-12T08:13:35Z

hw/arm/guest-socket.c

+    } else if ((guest_svcs_fds[retval] =
+#ifndef __APPLE__
+                                         accept4(guest_svcs_fds[sckt],
+                                         (struct sockaddr *) &addr,
+                                         &addrlen, SOCK_NONBLOCK)) < 0)
+#else
+                                         accept(guest_svcs_fds[sckt],
                                         (struct sockaddr *) &addr,
-                                         &addrlen)) < 0) {
+                                         &addrlen)) < 0)
+#endif
+    {
        retval = -1;
-        guest_svcs_errno = errno;
+        guest_svcs_errno = darwin_error(errno);
    } else {
+#ifndef __APPLE__
+        struct darwin_sockaddr_in darwin_addr = convert_sockaddr_to_darwin(addr);
+        cpu_memory_rw_debug(cpu, (target_ulong) g_addr, (uint8_t*) &darwin_addr,
+                                    sizeof(darwin_addr), 1);
+#endif


Let's use just accept, and copy the flags in the #ifndef __APPLE__ section.

aronsky · 2020-08-12T08:15:15Z

include/hw/arm/guest-services/socket.h

+#define SOCKADDR	    sockaddr
+#define SOCKADDR_IN	    sockaddr_in


No need for these defines. You can use darwin_sockaddr/darwin_sockaddr_in directly. If you don't want to redefine the structs in Darwin, you can use #define darwin_sockaddr sockaddr and #define darwin_sockaddr_in sockaddr_inwhen __APPLE__ is defined.

MCApollo · 2020-08-12T08:35:19Z

accept4:
Working on it- probably finish tomorrow. Do suggest a place to pull/set flags off the listening socket.
sockaddr macro:
I feel like that affects readability for future contributors, but can be still changed if wished.

That previous ADDR macro was there to avoid reductant/repeating cpu_rw code (if that’s a problem).

aronsky · 2020-08-12T08:58:02Z

you can set the flags right in the next block (under else), inside the #ifndef __APPLE__ block
I think if you use darwin_sockaddr in all the relevant spots, it's quite readable. It's clear that it's a sockaddr structure, and also that it's Darwin-specific.

aronsky · 2020-08-26T17:06:33Z

@MCApollo any progress with this?

Maroc-OS · 2020-08-28T01:23:40Z

bind is not working even on macOS after the latest changes, error 45, can't work on this no time, but just to let you know

MCApollo · 2020-08-28T03:35:42Z

Run it under lldb for me, this PR shouldn’t affect any macos parts at all.

Maroc-OS · 2020-08-28T04:03:55Z

Run it under lldb for me, this PR shouldn’t affect any macos parts at all.

i mean when kvm was implemented the tunneling stoped working so.
read from ''' Implementation-defined registers and user-space ''' here :

https://alephsecurity.com/2020/07/19/xnu-qemu-kvm/

aronsky · 2020-08-30T06:09:26Z

@Maroc-OS are you running on ARM with KVM, or is there an issue when using regular QEMU emulation? Because the KVM changes should not affect the latter.

Maroc-OS · 2020-09-01T01:07:10Z

@Maroc-OS are you running on ARM with KVM, or is there an issue when using regular QEMU emulation? Because the KVM changes should not affect the latter.

@aronsky yep regular QEMU emulation.

After updating to kvm commits it stopped working, and no ssh is there.

here is a little script am using adapted from @MCApollo :

#!/bin/sh

export XNU_SOURCES=$PWD/darwin-xnu
export KERNEL_SYMBOLS_FILE=$PWD/symbols.nm
export QEMU_DIR=$PWD/xnu-qemu-arm64/
export QEMU_TOOLS_DIR=$PWD/xnu-qemu-arm64-tools/
export NUM_BLOCK_DEVS=2
export KERNEL_CACHE=$PWD/kernelcache.release.n66.out
export DTB_FIRMWARE=$PWD/Firmware/all_flash/DeviceTree.n66ap.im4p.out
export DRIVER_FILENAME=$PWD/aleph_bdev_drv.bin
export IOS_DIR=$PWD
export HFS_MAIN=$PWD/hfs.main
export HFS_SEC=$PWD/hfs.sec


# Because there is no graceful shutdown of the system. mount & remount disk image
hdiutil attach -imagekey diskimage-class=CRawDiskImage ${HFS_MAIN}
hdiutil detach /Volumes/PeaceB16B92.arm64UpdateRamDisk
hdiutil attach -imagekey diskimage-class=CRawDiskImage ${HFS_SEC}
hdiutil detach /Volumes/PeaceB16B92.arm64UpdateRamDisk

# Update tree & Build the Custom Block Device Driver
cd ${QEMU_TOOLS_DIR}
git pull
cd ${IOS_DIR}

make -C ${QEMU_TOOLS_DIR}/aleph_bdev_drv
cp ${QEMU_TOOLS_DIR}/aleph_bdev_drv/bin/aleph_bdev_drv.bin ${DRIVER_FILENAME}

# Update tree & Build XNU QEMU for iOS
cd ${QEMU_DIR}
git pull

./configure \
--target-list=aarch64-softmmu \
--disable-capstone \
--disable-slirp \
--disable-pie \

export nproc=`sysctl -n hw.ncpu`
make -j${nproc}

cd ${IOS_DIR}

# Launch the kernel :)
xnu-qemu-arm64/aarch64-softmmu/qemu-system-aarch64 \
-M iPhone6splus-n66-s8000,kernel-filename=${KERNEL_CACHE},dtb-filename=${DTB_FIRMWARE},driver-filename=${DRIVER_FILENAME},qc-file-0-filename=${HFS_MAIN},qc-file-1-filename=${HFS_SEC},kern-cmd-args="debug=0x8 kextlog=0xfff cpus=1 rd=disk0 serial=2",xnu-ramfb=off \
-cpu max \
-m 6G \
-serial mon:stdio \
-vga std \

resulting crash of "MyUnixSocket" which is responsible of loading the (signed/unsigned both tested) tcp-tunnel binary, trying to launch it manually resulting "error 45" at binding

Added Linux support to tcp tunneling

0839660

MCApollo mentioned this pull request Aug 6, 2020

tcp-tunnel: botch Linux support alephsecurity/xnu-qemu-arm64-tools#15

Closed

guest-socket: clean up addrlen's

59a223a

guest-socket: reformat code; handle errno's on host side

1377261

aronsky requested changes Aug 11, 2020

View reviewed changes

guest-socket: Improved readability and flow

df08c4c

aronsky requested changes Aug 12, 2020

View reviewed changes

guest-socket: inherit flags from fd

1f63f61

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Added Linux support to qc_handle_{accept,bind,connect} functions. #23

Added Linux support to qc_handle_{accept,bind,connect} functions. #23

MCApollo commented Aug 6, 2020 •

edited

Loading

sickcodes commented Aug 7, 2020 •

edited

Loading

MCApollo commented Aug 7, 2020

sickcodes commented Aug 7, 2020 •

edited

Loading

sickcodes commented Aug 7, 2020

sickcodes commented Aug 7, 2020 •

edited

Loading

sickcodes commented Aug 7, 2020 •

edited

Loading

sickcodes commented Aug 8, 2020

aronsky commented Aug 9, 2020

MCApollo commented Aug 9, 2020

aronsky Aug 11, 2020

aronsky Aug 11, 2020

aronsky Aug 11, 2020

aronsky Aug 11, 2020

aronsky Aug 11, 2020

aronsky Aug 11, 2020

aronsky Aug 11, 2020

aronsky Aug 11, 2020

aronsky Aug 11, 2020

aronsky Aug 11, 2020

MCApollo commented Aug 11, 2020 •

edited

Loading

aronsky commented Aug 11, 2020

MCApollo commented Aug 11, 2020 •

edited

Loading

aronsky left a comment

aronsky Aug 12, 2020

aronsky Aug 12, 2020

MCApollo commented Aug 12, 2020

aronsky commented Aug 12, 2020

aronsky commented Aug 26, 2020

Maroc-OS commented Aug 28, 2020

MCApollo commented Aug 28, 2020

Maroc-OS commented Aug 28, 2020 •

edited

Loading

aronsky commented Aug 30, 2020

Maroc-OS commented Sep 1, 2020 •

edited

Loading

	static struct sockaddr_in convert_sockaddr_to(S_SOCKADDR_IN from) {
	static struct sockaddr_in convert_sockaddr_darwin_to_gnu(darwin_sockaddr_in from) {

Added Linux support to qc_handle_{accept,bind,connect} functions. #23

Are you sure you want to change the base?

Added Linux support to qc_handle_{accept,bind,connect} functions. #23

Conversation

MCApollo commented Aug 6, 2020 • edited Loading

sickcodes commented Aug 7, 2020 • edited Loading

MCApollo commented Aug 7, 2020

sickcodes commented Aug 7, 2020 • edited Loading

sickcodes commented Aug 7, 2020

sickcodes commented Aug 7, 2020 • edited Loading

sickcodes commented Aug 7, 2020 • edited Loading

sickcodes commented Aug 8, 2020

aronsky commented Aug 9, 2020

MCApollo commented Aug 9, 2020

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

MCApollo commented Aug 11, 2020 • edited Loading

aronsky commented Aug 11, 2020

MCApollo commented Aug 11, 2020 • edited Loading

aronsky left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

MCApollo commented Aug 12, 2020

aronsky commented Aug 12, 2020

aronsky commented Aug 26, 2020

Maroc-OS commented Aug 28, 2020

MCApollo commented Aug 28, 2020

Maroc-OS commented Aug 28, 2020 • edited Loading

aronsky commented Aug 30, 2020

Maroc-OS commented Sep 1, 2020 • edited Loading

MCApollo commented Aug 6, 2020 •

edited

Loading

sickcodes commented Aug 7, 2020 •

edited

Loading

sickcodes commented Aug 7, 2020 •

edited

Loading

sickcodes commented Aug 7, 2020 •

edited

Loading

sickcodes commented Aug 7, 2020 •

edited

Loading

MCApollo commented Aug 11, 2020 •

edited

Loading

MCApollo commented Aug 11, 2020 •

edited

Loading

Maroc-OS commented Aug 28, 2020 •

edited

Loading

Maroc-OS commented Sep 1, 2020 •

edited

Loading