# After convert 8.3-\>9.1, hexrays stack type are lost

**URL:** <https://community.hex-rays.com/t/after-convert-8-3-9-1-hexrays-stack-type-are-lost/245>\
**Category:** Bugs\
**Created:** [March 16, 2025, 4:00pm UTC](https://community.hex-rays.com/t/after-convert-8-3-9-1-hexrays-stack-type-are-lost/245 "2025-03-16T16:00:40Z")\
**Posts on this page:** 6\
**Page:** 1

<div class="post-metadata">

**Author:** ![revs](https://avatars.discourse-cdn.com/v4/letter/r/dc4da7/32.png) [@revs](https://community.hex-rays.com/u/revs)\
**Post date:** [March 16, 2025, 4:00pm UTC](https://community.hex-rays.com/t/after-convert-8-3-9-1-hexrays-stack-type-are-lost/245/1 "2025-03-16T16:00:40Z")

</div>

idb 8.3  
in hexrays i set type of var std\_string (this my struct size of 0x20)  
but when press ENTER in hexrays view on this name of type  
i see

> -0000000000000098 var\_98 ymmword ?  
> -0000000000000078 var\_78 ymmword ?  
> -0000000000000058 var\_58 ymmword ?  
> -0000000000000038 var\_38 ymmword ?

in hexrays 8.3 all ok

but after convert idb 8.3 to idb 9.1  
both types on stack and in hexrays became as \_\_m256

> -0000000000000098 \_\_m256 var\_98;  
> -0000000000000078 \_\_m256 var\_78;  
> -0000000000000058 \_\_m256 var\_58;  
> -0000000000000038 \_\_m256 var\_38;

after conversion, hexrays stack types were lost  
but not all  
but only some  
I suppose only those that were not tied to the ida stack

you yourself made it so that in the flight you can set the type separately from the ida stack

this functionality is poorly developed in the conversion idb

---

<div class="post-metadata">

**Author:** ![igor](https://sea2.discourse-cdn.com/flex002/user_avatar/community.hex-rays.com/igor/32/35_2.png) [@igor](https://community.hex-rays.com/u/igor)\
**Post date:** [March 17, 2025, 2:15pm UTC](https://community.hex-rays.com/t/after-convert-8-3-9-1-hexrays-stack-type-are-lost/245/2 "2025-03-17T14:15:56Z")

</div>

IDA 9 does not support legacy asm-style structures, so they have to be converted to new type system during upgrade. The decompiler-specific type storage is no longer used so in case of conflicts stack frame members are taken from IDA.

---

<div class="post-metadata">

**Author:** ![revs](https://avatars.discourse-cdn.com/v4/letter/r/dc4da7/32.png) [@revs](https://community.hex-rays.com/u/revs)\
**Post date:** [March 17, 2025, 3:43pm UTC](https://community.hex-rays.com/t/after-convert-8-3-9-1-hexrays-stack-type-are-lost/245/3 "2025-03-17T15:43:24Z")

</div>

so in IDA 9 all hexrays stack vars = ida stack vars  
rights ?  
its interesting how about cases when

> hexrays stack vars != ida stack vars

as far as I remember it was introduced for complex cases  
and I had such cases  
unfortunately I can’t remember them right now  
to check what will happen in pseudocode

---

<div class="post-metadata">

**Author:** ![igor](https://sea2.discourse-cdn.com/flex002/user_avatar/community.hex-rays.com/igor/32/35_2.png) [@igor](https://community.hex-rays.com/u/igor)\
**Post date:** [March 19, 2025, 11:25am UTC](https://community.hex-rays.com/t/after-convert-8-3-9-1-hexrays-stack-type-are-lost/245/4 "2025-03-19T11:25:15Z")

</div>

There are still cases where the decompiler keeps additional information about local stack variables, such as split or mapped variables, but in most cases the stack frame type info is now shared between IDA and the decompiler.

---

<div class="post-metadata">

**Author:** ![revs](https://avatars.discourse-cdn.com/v4/letter/r/dc4da7/32.png) [@revs](https://community.hex-rays.com/u/revs)\
**Post date:** [March 19, 2025, 12:26pm UTC](https://community.hex-rays.com/t/after-convert-8-3-9-1-hexrays-stack-type-are-lost/245/5 "2025-03-19T12:26:28Z")

</div>

does this have anything to do with it?

```auto
Type info leak has been detected and fixed (refcnt=1; idx=50584)
Type info leak has been detected and fixed (refcnt=1; idx=50637)
Type info leak has been detected and fixed (refcnt=1; idx=50640)
Type info leak has been detected and fixed (refcnt=1; idx=50653)
Type info leak has been detected and fixed (refcnt=1; idx=50779)
Type info leak has been detected and fixed (refcnt=1; idx=50784)

```

---

<div class="post-metadata">

**Author:** ![igor](https://sea2.discourse-cdn.com/flex002/user_avatar/community.hex-rays.com/igor/32/35_2.png) [@igor](https://community.hex-rays.com/u/igor)\
**Post date:** [March 19, 2025, 1:58pm UTC](https://community.hex-rays.com/t/after-convert-8-3-9-1-hexrays-stack-type-are-lost/245/6 "2025-03-19T13:58:33Z")

</div>

> [@revs](#):
>
> does this have anything to do with it?

Probably not; this message usually indicates an issue in IDA or some plugin/script.
