🐛 EXC_BAD_ACCESS from closure in updateOfferingsCache

[esamelson]

Describe the bug

I'm seeing this crash reported from a few devices in production after switching to RevenueCat recently. It looks like this code was touched recently, so it's possible it's fixed in a newer version of the SDK (I have 5.43.0 in production currently) but wanted to report anyway as I didn't see any other reports of this specific crash.

I'm using the React Native SDK and doing nothing with offerings other than calling Purchases.getOfferings.

Platform

iOS

SDK version

5.43.0

SDK integration method

CocoaPods

StoreKit version

StoreKit 2 (default on versions >=5.0.0)

OS version

iOS 18.0

Xcode version

16.1

Device and/or simulator

Device

Environment

Production

How widespread is the issue

Only specific devices (but persistent)

Debug logs

## Exception

### Exception
**Type:** EXC_BAD_ACCESS
**Value:** Exception 1, Code 1, Subcode 4 >
KERN_INVALID_ADDRESS at 0x4.

#### Stacktrace


 closure in Purchases.updateOfferingsCache in unknown file [Line null] (In app)
 swift::runJobInEstablishedExecutorContext in unknown file [Line null] (Not in app)

Steps to reproduce

(using React Native)
Purchases.configure();
...
await Purchases.getOfferings();

note: I have not reproduced this myself, calls are reconstructed from crash logs

Other information

Full information/diagnosis from Sentry

# EXC_BAD_ACCESS: Exception 1, Code 1, Subcode 4 >

## Issue Summary
EXC_BAD_ACCESS in Purchases.updateOfferingsCache
**What's wrong:** EXC_BAD_ACCESS (KERN_INVALID_ADDRESS) in **Purchases.updateOfferingsCache** closure, likely due to **null pointer dereference**.
**Possible cause:** The crash may stem from an **uninitialized or deallocated object** being accessed within the `updateOfferingsCache` closure, possibly related to network responses.

## Tags

- **device:** iPad13,18
- **device.class:** high
- **device.family:** iOS
- **environment:** release
- **event.environment:** native
- **event.origin:** ios
- **handled:** no
- **level:** fatal
- **mechanism:** mach
- **os:** iOS 18.0
- **os.build:** 22A5307f
- **os.name:** iOS
- **os.rooted:** no

## Exception

### Exception 1
**Type:** EXC_BAD_ACCESS
**Value:** Exception 1, Code 1, Subcode 4 >
KERN_INVALID_ADDRESS at 0x4.

#### Stacktrace


 closure in Purchases.updateOfferingsCache in unknown file [Line null] (In app)
 swift::runJobInEstablishedExecutorContext in unknown file [Line null] (Not in app)

Additional context

[RCGitBot]

👀 We've just linked this issue to our internal tracker and notified the team. Thank you for reporting, we're checking this out!

[JayShortway]

Hi @esamelson, thanks for reporting this! Does Sentry provide information about the app state such as foreground/background? Also, does the stack trace contain any more lines, or just those 2? We'll investigate and get back to you.

[JayShortway]

Hi @esamelson, one more question: could you list all the iOS versions you're seeing this on? Thanks in advance!

[esamelson]

Thanks @JayShortway -- so far we are only seeing this on iOS 18.0, but it's a small sample size. The app is in the foreground when this happens. Looking more closely I don't think Purchases.getOfferings() has even been called, and it's being triggered just from configuring RevenueCat with an appUserID.

The stacktrace doesn't have any additional lines beyond what I initially posted:

Exception Type: EXC_BAD_ACCESS (SIGSEGV)
Exception Codes: SEGV_NOOP at 0x0000000000000004
Crashed Thread: 3

Application Specific Information:
Exception 1, Code 1, Subcode 4 >
KERN_INVALID_ADDRESS at 0x4.

Thread 3 Crashed:
0   <app-name>                      0x204813944         closure in Purchases.updateOfferingsCache
1   libswift_Concurrency.dylib      0x33180c8d0         swift::runJobInEstablishedExecutorContext

[ajpallares]

Hi @esamelson! Thank you for the report!

I wasn't able to reproduce the crash, not even running in iOS 18.0. I suspect it mat be a bug in the Swift Concurrency runtime, very likely fixed in later versions of iOS, and probably made worse by some redundancies in tasks and task groups creation in our codebase.

Even though we don't have an explicit fix for this, I did just merge this PR #5847 simplifying the implementation of Purchases.updateOfferingsCache, where the crash seems to originate. If the crashes were due to that complexity in our codebase, this change should improve reduce the occurrences.

Next week we'll make a new release of react-native-purchases containing this change. Please, do let us know if you keep getting the same crash report using next week's version of the SDK.

Thanks again!

[esamelson]

Thanks very much for the speedy diagnosis and update, @ajpallares ! We will try out this change once it's released and let you know if we continue to see the issue.