Join Us and become a Member for a Verified Badge on Discord to access private areas with the latest PS4 FPKGs.
Category PS4 Guides and Tutorials       Thread starter PSXHAX       Start date Aug 11, 2020 at 12:49 AM       5,768       17            
Proceeding his PS4 ROP 8CC Port, Bad_Hoist 6.72 PS4 Webkit Exploit Port and preliminary Checklist to Port the Toolchain, today PlayStation 4 developer sleirsgoevy updated his Github repository with a Bad_Hoist PS4 Toolchain Porting Guide for Unsupported Firmware for those interested in learning how it's done. šŸŽ“

:arrow: In addition, Peeter (aka PaulJenkin) has also forked the Github repository with a more 'noob friendly' Bad_Hoist GIT Fork and Porting Guide as well! :tup:

From the, to quote: Porting guide

This document will guide you through the process of getting this toolchain up and running on an unsupported firmware. The only implicit assumption made here is that the core exploit (exploit.js) does run on the target firmware. It shouldn't be too difficult to swap that for another userland exploit, but it's not discussed here.

This is more-or-less what I did for 6.51 (I didn't have any userspace dumps back then). 6.72 happened to be "compatible enough", so no more porting was required.

1. Find GOT offset

Dump WebKit using python3 bad_hoist/memserver/ -1 (this should never fail). Then disassemble the dumped binary, extract all CALL imm32 instructions, and sort by the target address:
objdump -D bad_hoist/dumps/webkit.elf | grep 'call[^%]*$' | cut -d "$(printf '\t')" -f 3- | sort | uniq | less
You'll see something like this in the dump:
callq  ffffffffff0b9c41 <__bss_start+0xfffffffffd7aec41>
callq  ffffffffff2bf176 <__bss_start+0xfffffffffd9b4176>
callq  ffffffffff65df97 <__bss_start+0xfffffffffdd52f97>
callq  ffffffffff667308 <__bss_start+0xfffffffffdd5c308> <-- probably the start of the GOT
callq  ffffffffff667318 <__bss_start+0xfffffffffdd5c318>
callq  ffffffffff667338 <__bss_start+0xfffffffffdd5c338>
callq  ffffffffff667348 <__bss_start+0xfffffffffdd5c348>
callq  ffffffffff667358 <__bss_start+0xfffffffffdd5c358>
callq  ffffffffff667368 <__bss_start+0xfffffffffdd5c368>
callq  ffffffffff667378 <__bss_start+0xfffffffffdd5c378>
callq  ffffffffff667388 <__bss_start+0xfffffffffdd5c388>
callq  ffffffffff667408 <__bss_start+0xfffffffffdd5c408>
callq  ffffffffff667418 <__bss_start+0xfffffffffdd5c418>
Here 0xffffffffff667308 is the first of a few hundreds of similarly looking entries. Probably it's the start of the GOT (or PLT, don't care much about the terminology). This means that the GOT is located (1 << 64) - 0xffffffffff667308 bytes below the leaked virtual method.

Now insert this offset into bad_hoist/dumpers/dump_got.js at line 11, and load a page that loads exploit.js, helpers.js, malloc.js, dumpers/dump_got.js. print function should be somehow routed to your dev PC.

If the script warns about "decrement!", this means that there is one more GOT entry that didn't get referenced. Increment the offset by 16.

2. Find GOT entries corresponding to the required modules

Run the dump_got.js script from above, this time with the correct offset. You'll get a long list of pointers, some of which are 0xeffffffe00000000, and some others are actual code pointers. Save this output to a file (it will be useful later) and sort the pointers by their value.

Here for each module are a few dozens of pointers pointing into it. If two consecutive pointers differ by hundreds of megabytes, then most probably the previous module ended here and a new one just started. Example:
<-- here
In a 6.72 dump, 5 modules can be obviously separated this way. It turns out that both libc and libkernel are amongst them.

3. Dump modules

Once you know a "base address" (not a base address actually, but rather address of some random function inside the module) you can try dumping the module. Go to the file you saved at step 2 (a fresh dump won't be useful because of userland ASLR) and search for the (copy-pasted from sorted) lowest pointer of a specific module. If the pointer is at line X+1, then this module can be dumped by running python3 bad_hoist/memserver/ X.

(Note: before dumping, replace the GOT offset in with the offset from dump_got.js)

Now dump each of these modules and check if it's either libc or libkernel:
  • libc is the module that contains a pivot() gadget (mov rsp, [rdi+0x38] ; pop rdi ; ret). The gadget is located at the end of an easily recognizable loadall() function.
  • libkernel contains easily recognizable syscall wrappers. Search for syscall in the disassembly; if what you see looks like a syscall wrapper, you've just found libkernel.
Note: as the base address used is not the real base address, some important function may be located below that address. To address this case, supports a index,offset notation as its argument (see bad_hoist/memserver/Makefile). E.g. to get pthread_create into the dump I had to pass 705,-0x10000 instead of just 705, so that dumping starts 64KB lower than the pointer.

Once you've verified that you'd dumped the right modules, fix bad_hoist/memserver/Makefile with the correct offsets and run the dumping scripts in "production mode":
bad_hoist/ $ make clean
bad_hoist/ $ make
4. Fix hardcoded offsets in rop/rop.js
  1. Line 16. Insert the correct GOT offset.
  2. Lines 33-34. Insert the correct libc & libkernel bases. The numbers are the same as in Makefile.
  3. Lines 35-43:
3.1. loadall is the function in libc that contains the pivot() gadget.

3.2. saveall is a twin function of loadall that saves registers instead of restoring them.

3.3. setjmp and longjmp are not used, as loadall and saveall play the same role here. You can just strip these out.

3.4. pivot is the pivot() gadget, see above.

3.5. infloop is a EB FE (.loop\njmp loop) gadget, not used in the end exploit but is useful for debugging ropchains.

3.6. jop_frame is a JOP gadget that reads as push rbp ; mov rbp, rsp ; mov rax, [rdi] ; call qword [rax]. It is used for cleanly returning from subthreads, see ps4-rop-8cc/librop/extcall.{c,h}.

3.7. get_errno_addr is a function that returns the address of the errno(3) variable for the current thread. To find this address, follow the jb jump after any syscall instruction, then follow an indirect jump, and you will see a CALL to this function.

3.8. pthread_create is the only function in libkernel that calls the thr_new syscall wrapper. First find the wrapper by searching for the thr_new syscall number, then find a call to this wrapper, then search backward for a function prologue.
Bad_Hoist PS4 Toolchain Porting Guide for Unsupported Firmware by Sleirsgoevy.jpg



i don't understand at all what is this for but i know it is important to the scene... thank you so much for all your hardwork!!


I have ported this jailbreak to 6.20, right now we are getting kernel panic every time we invoke the jailbreak, will improve the same in up coming days.

Attaching the self host files here: 6.20 JB (0.27 MB). Also created few more scripts which will ease this process will post them all to github in few days.

Credits to sleirsgoevy and LightningMods for helping me on this.


@RF0oDxM0Dz This is not a web kit, this is back port of sleirsgoevy jb to 6.20

It servers only below purpose:

1. People who are stuck in 6.20 with BD broken and cant update will be able to use this as their JB
2. I have created additional python script to automate manual work involved in porting
3. This will be helpful, while porting this JB to 7.02, when a web kit for 7.02 is released
4. Will be sending pull request to update the current porting guide, So it help out the beginners

I have Forked sleirsgoevy badhoist repo and created more noob friendly repo and guide under GIT Repo, GIT Guide, added below additional script which will help everyone to do this very easily:
  1. Added php script which dumps the GOT offset in synchronised order
  2. Added py script for finding the GOT base address
  3. Added py script for finding the base address jumps
  4. Added py script for finding the address for rop
Note: i have also submitted this change to sleirsgoevy as a pull request.
Recent Articles
PlayStation 5 Model CFI-1018A Console Images with PS5 Accessories via FCC
Thus far we've seen the official DualSense Wireless Controller Images, official PlayStation 5 Images, some Prototype Images and with the PS5 Price and Release Date known this weekend geronimo_73...
Sony CEO Confirms PS5 Won't Natively Support PS1, PS2 nor PS3 Games
In contrast to Microsoft's upcoming Xbox Series X console that will also play original Xbox, Xbox 360 and Xbox One games at launch, Sony CEO Jim Ryan confirmed in a Famitsu Interview following the...
Call of Duty: Black Ops Cold War PS4 Alpha Live September 18-20th
Proceeding the Call of Duty: Black Ops Cold War PS4 and Call of Duty: Black Ops Cold War PS4 Multiplayer trailers, starting today Activision announced that the Call of Duty: Black Ops Cold War PS4...
Demon's Souls PS5 Gameplay Trailer and More PlayStation 5 Videos!
As a massive fan of the original King's Field series on PSX and PS2 from the Japanese video game company From Software, this is one franchise I don't mind seeing milked unlike so many other...